Schizophrenia Research Forum - A Catalyst for Creative Thinking

The Shifting Diagnostic Sands of Psychosis

30 June 2011. A new study throws a bright light on the difficulty of diagnosing psychotic disorders. Published online June 15 in the American Journal of Psychiatry, the study found that the diagnosis initially given to 50 percent of study participants who had been admitted to the hospital for psychosis changed at least once during the ensuing 10 years. These shifts mainly reflected changing symptoms over the course of life, rather than misdiagnoses.

"Everyone knows that symptoms and illness course and life circumstances change over time. But a significant message of this paper is that these factors can alter the central diagnosis, which typically is not revisited," said Evelyn Bromet of Stony Brook University School of Medicine, who led the study. "When studying adults, everybody assumes that they've grown up and matured and they're not moving targets anymore. But it's not true."

The findings mark the most recent installment of a longitudinal study comparing diagnoses given at first admission for psychosis, six months, two years, and 10 years afterward. The four time points reveal illness trajectories that fluctuate across five broad diagnostic categories, which included schizophrenia, bipolar disorder, major depression, substance-induced psychosis, and other psychotic conditions. Most changes consisted of shifts from a non-schizophrenia category at baseline to schizophrenia later in life, with one-third of this group getting the schizophrenia diagnosis after year two.

"I think the assumption was that we sorted the diagnosis out by two-year follow-up," Bromet told SRF. "But we found a lot can change between two and 10 years."

One decade later
When the study began in the late 1980s, Bromet hadn't expected diagnoses to change much, especially after the six-month follow-up timepoint. But when contacting people later, it became clear that symptoms had often changed enough to warrant a new diagnosis. "When we started, longitudinal studies did not re-diagnose people," she said.

For the new study, diagnoses were made at 10 years for 470 people who had been systematically assessed at their first hospital admission for psychosis, six months later, and two years later. The same procedure was used to make a diagnosis at each time point: a mental health professional interviewed each person face to face, and at least four psychiatrists used these and previous assessments, medical records, and interviews with family members to settle on a diagnosis. However, the psychiatrists remained blind to previous research diagnoses.

The data revealed a major shift toward schizophrenia diagnoses. Of participants receiving a non-schizophrenia diagnosis at baseline, 32 percent eventually received a schizophrenia diagnosis at year 10. Of those initially receiving a non-bipolar disorder diagnosis, 10 percent ultimately received a bipolar diagnosis at year 10.

The detailed distribution at baseline was 30 percent schizophrenia, 21 percent bipolar disorder, 17 percent major depression, 2 percent substance-induced psychosis, and 28 percent other psychoses. At year 10, this became 50 percent, 24 percent, 11 percent, 7 percent, and 8 percent, respectively. The increase in schizophrenia and bipolar diagnoses, coupled with the decrease in the "other" category, suggests that these illnesses can take time to make themselves apparent.

Diagnostic stability on the horizon?
Does the 10-year diagnosis reflect a stabilizing illness course, or yet another fluctuation in symptoms? Consistent with the idea that diagnostic stability can eventually be achieved for some cases, baseline diagnoses of schizophrenia and bipolar disorder largely held true at year 10, with 90 percent of initial schizophrenia diagnoses and 78 percent of initial bipolar disorder diagnoses retained.

Still, 50 percent of study participants received a different diagnosis at least once during the follow-up period, and their trajectories differed: some ended up with the same initial diagnosis with which they began, whereas others received a wholly different diagnosis. To explore whether these fluctuations would dampen with time and stabilize upon one diagnosis, Bromet plans to reassess the cohort at 20 years.

The researchers also tracked down the changes in the clinical picture that spurred the diagnostic shifts. Whereas poorer functioning and greater negative and psychotic symptoms predicted a shift to schizophrenia, better functioning and lower negative and depressive symptoms predicted a shift to bipolar disorder.

If clearer diagnostic pictures emerge with time, the results are bound to concern clinicians and researchers alike. Shifting diagnoses not only complicate treatment strategies, but they also muddy the waters for genetic and postmortem research, which depends on a reliable diagnosis as a starting point.

"It's really only fair, if you're treating somebody, to reconsider the diagnosis," Bromet said. "It doesn't mean a diagnosis made 10 years ago wasn't valid then. It just means that it may or may not be valid today."—Michele Solis.

Bromet EJ, Kotov R, Fochtmann LJ, Carlson GA, Tanenberg-Karant M, Ruggero C, Chang SW. Diagnostic Shifts During the Decade Following First Admission for Psychosis. Am J Psychiatry. 2011 Jun 15. Abstract

Comments on News and Primary Papers

Primary Papers: Diagnostic Shifts During the Decade Following First Admission for Psychosis.

Comment by:  Patrick McGorry
Submitted 11 July 2011
Posted 11 July 2011

Rather than showing "misclassification" early on, Bromet et al. really show how inadequate these concepts, especially schizophrenia, are for early diagnosis and treatment. We need a model which links stage of illness to treatment selection and allows people with a need for care (often months or years before the current diagnostic approach really makes sense) to receive the help they need. Our Clinical Staging Model is an attempt to work towards this (McGorry et al., 2006).


McGorry PD, Hickie IB, Yung AR, Pantelis C, Jackson HJ. 2006. Clinical staging of psychiatric disorders: a heuristic framework for choosing earlier, safer and more effective interventions. The Australian and New Zealand Journal of Psychiatry 40(8): 616-622.

McGorry PD, Nelson B, Goldstone S, Yung R. 2010. Clinical staging: a heuristic and practical strategy for new research and better health and social outcomes for psychotic and related mood disorders. Canadian Journal of Psychiatry 55(8): 486-497.

View all comments by Patrick McGorry

Primary Papers: Diagnostic Shifts During the Decade Following First Admission for Psychosis.

Comment by:  Lawrence YangEzra Susser (SRF Advisor)
Submitted 27 July 2011
Posted 27 July 2011

We believe that the paper by Evelyn Bromet and colleagues will come to be viewed as a classic in psychiatric research. It addresses a fundamental problem that has long been recognized, but nevertheless is often overlooked in published papers, and sometimes in clinical practice and mental health service policies. One of the reasons that has made it easier to overlook this problem is that it has not been directly addressed with high-quality empirical data. This study changes that.

Briefly, the problem is that psychiatric diagnoses are often made based on the state of an individual at a given point in time. Yet we know that individual states change over time, and that longitudinal information (not always available) will generate a more valid diagnosis. As noted in their paper, this was recognized in the classic paper of Robins and Guze (1970), but it has also been emphasized by others since then. A paper by Robert Spitzer suggested that, while we have no “gold standard” in psychiatry, we could use a “LEAD” (“Longitudinal, Expert, and All Data”) standard, a central feature of which was using longitudinal data to help verify diagnoses (Spitzer, 1983).

Dr. Bromet’s paper now reveals that, in a sample of individuals with psychoses, ascertained at first hospital admission, with carefully made diagnoses over several time points up to 10-year follow-up, the shifts in diagnosis were substantial. They were to some extent predictable, but not enough to warrant assuming the validity of a diagnosis based on evaluation at a single time point early in the course of the illness. Clearly, we can no longer afford to overlook the difference between diagnoses of psychotic disorders that are made based on cross-sectional versus longitudinal information. (Arguably a diagnosis of schizophrenia could be an exception to this rule, since individuals were very likely to retain this diagnosis if it had been made at the time of entry to the study.)

While these findings pertain to a treated sample in a high-income country (the U.S.), the implications are global, and there are a number of ways in which this question could be examined from other vantage points in non-Western settings. For example, consider a recent study of mental disorders in a representative community sample in China (Phillips et al., 2009). This study was large enough to ascertain many individuals with psychoses and (unusual for community surveys) used a clinician-administered interview to establish their diagnoses (SCID; Spitzer et al., 1992). Among the individuals diagnosed as having schizophrenia, about 25 percent were untreated, and another 25 percent had been minimally treated. This study did not follow individuals over time, but it illustrates the potential and importance of doing so for (previously) untreated samples identified in a non-Western setting (Ran et al., 2001). Examination of this group might provide a different perspective by which to study long-term diagnostic stability of psychotic disorders.


Spitzer, R.L. (1983). Psychiatric diagnosis: are clinicians still necessary? Comprehensive Psychiatry, 24, 399-411. Abstract

Phillips MR, Zhang J, Shi Q, et al. Prevalence, treatment, and associated disability of mental disorders in four provinces in China during 2001-05: an epidemiological survey. The Lancet. 2009;373(9680):2041-2053. Abstract

Spitzer RL, Williams JBW, Gibbon M, First MB. The Structured Clinical Interview for DSM-III-R (SCID) I: History, Rationale, and Description. Arch Gen Psychiatry. 1992;49(8):624-629. Abstract

Ran M et al. Natural course of schizophrenia: 2-year follow-up study in a rural Chinese community. The British Journal of Psychiatry. 2001;178(2):154-158. Abstract

View all comments by Lawrence Yang
View all comments by Ezra Susser

Primary Papers: Diagnostic Shifts During the Decade Following First Admission for Psychosis.

Comment by:  John Strauss
Submitted 10 August 2011
Posted 10 August 2011

This is an important and valuable piece of work. The attention to the reliability of ratings and diagnosis is well done, and achieving an 80 percent follow-up over 10 years in a U.S. sample is impressive indeed. I would be interested to know how they managed that. Documenting diagnostic consistency and shifts provides important data for our understanding of: 1) diagnostic processes and 2) the course of psychiatric disorder. I put “diagnostic processes” first because we have such an unscientific tendency to treat diagnoses like real “things,” as though we were talking about something like a Toyota Corolla, rather than as complex interactions occurring in social contexts. Questions crucial to the diagnostic process include, among other things, how and where the patient is seen, the training of the person giving the diagnosis, the diagnostic system in use, and (sometimes finally), the characteristics of the person being given the label. Although some of these concerns are impressively considered in this study, one important assumption discussed below is not.

Diagnoses are a crucial part of our field; they provide handy labels that can be useful for treatment, training and research. Or they can provide misleading reifications that indicate that something is “a disease” when in fact we know much less about that thing than having a diagnostic label would indicate. Knowing about changes and constancies in diagnoses may tell us about changes or lack thereof in patients. They may also tell us about implicit erroneous thinking in the person making the diagnosis. A joke among many consumers with severe psychiatric problems is that “They used to call me a schizophrenic, but when I got better they changed me to bipolar.” That could indicate that the person’s clinical state changed in a diagnostically relevant way, or it could indicate that the person making the diagnosis was influenced by the now disproved belief that people with schizophrenia never get better. The person making the diagnosis might assume that since this person improved, he or she must have been bipolar. That would be the changing of a diagnosis to fit a disproved belief of the person making the diagnosis. Or conversely, and one sees this on medical wards and even more in outpatient clinics, when a diagnosis is not changed, that might be because the person making the diagnosis is too harried (or heaven forfend, too lazy) to reevaluate the patient and that it’s just easier to write on the chart what has been written there before.

Another concern in the field of psychiatric diagnosis is the limits in our knowledge about the nature of psychiatric disorders, and here is where one major conclusion of this report is questionable. There are a lot of dicta these days running around as though they were definitive psychiatric truths, dicta like “psychiatric disorders are brain diseases” or “psychiatric illnesses are illnesses like any other." I am not here saying that these statements might not be true, but we really don’t know that they are or even exactly what they imply. As Robin Murray has noted, for example, the concept of “illnesses like any other” is rather bizarre since illnesses are not by any means all alike. Hypertension is too much of something, blood pressure, that at normal levels you'd better have if you want to be alive; pneumococcal pneumonia is entirely different; and, of course, a broken arm is different from both of these. In terms of a major conclusion of this report, the statement by Bromet et al. that “First-admission patients with psychotic disorders run the risk of being misclassified at early stages in the illness course,” may be seriously misleading. When a diagnosis changes, it could mean not that one or the other of the diagnoses was the correct one (and hence, that the disorders themselves don’t change, that you can only have one disorder and it’s just the question of getting it right), but it could mean that psychiatric disorders are not like that, that such a notion of disorder is incorrect, and that, on the contrary, one of our diagnostic types could fade into another over time. It is not rare, after all, if you know a patient over an extended period that you see what was an obsessional problem become a schizophrenic one or vice versa.

We don’t have to make either/or decisions about such possibilities—only recognize that our level of knowledge about psychiatric problems has serious limits. We can venture working hypotheses, which is better than assuming the truth of often implicit beliefs. Yes, diagnosis is very important, but it doesn’t necessarily mean that the diagnoses we have necessarily correspond to real unchanging things. The level of knowledge in psychiatry, as much as progress has been made, falls far short of allowing us to be sure we have the definitive story. Beware of nouns (such as diagnoses) in psychiatry! They may be indicating truths—or they may not. Very often, they reflect the best available working hypotheses about reality—no more and no less.

In my opinion, there are a lot of choices we do not have to make at this point in the trajectory of psychiatry. Specifically in regard to diagnosis, we do not have to decide whether or not to make a diagnosis, and then if we do, to treat it like a thing that couldn’t evolve into another thing. At this point, a psychiatric diagnosis is a helpful working hypothesis, not a definitive, all-revealing knowledge. I totally agree with the final conclusion of Bromet et al. for which their report provides important support: “Diagnosis should be reassessed at all follow-up points.” But I do not believe that their earlier conclusion is necessarily correct: that the first diagnosis can be a misclassification, and thus are implying that later diagnoses will identify the “real thing.” It seems highly likely to me that the disorders themselves can change, evolve, and resolve. We are not sure that a person with schizophrenia will always have schizophrenia. Is the last diagnosis always the “real” diagnosis, and are earlier different ones “misclassifications”? I don’t think we know that.

View all comments by John Strauss

Comments on Related News

Related News: Data Support Kraepelinian Boundary Between Psychotic Disorders

Comment by:  Irving GottesmanAksel Bertelsen
Submitted 23 October 2013
Posted 23 October 2013

Invigorating intellectual and heuristic debate in this Forum is kept alive by the challenging and informed summary of Kotov et al. by Michele Solis. The nagging problem of the status of schizoaffective disorder cannot be concluded by the evidence in hand from this study or others that are more biologically and genetically informed (e.g., B-SNIP data) because none are dispositive, to borrow a term from the lawyers. We applaud Kendler’s erudite and friendly dissection of Kotov et al. (Kendler, 2013) and concur with his conclusion that it would be premature to eliminate the Kraepelinian dichotomy. After all, the Alte Meister did not have access to GWAS or to DTI data from probands and their relatives, and ENCODE (Maurano et al., 2012) could not have been envisioned, either. We hope to supplement the SRF discussion with our twin (Cardno et al., 2012) and Scandinavian experiences (Bertelsen and Gottesman, 1995; Laursen et al., 2005; Gottesman et al., 2010; Lichtenstein et al., 2009). The last have cautioned against the tyranny of technology, while a British curmudgeon with a 2002 Nobel Prize, Sydney Brenner, has reminded us that one person’s junk is another’s treasure—the real task being how to organize data so that they yield knowledge.

First, we must compliment Kotov et al. for accomplishing the daunting task of successfully following up their U.S. cohort with 10 years of data. True, Manfred Bleuler completed an exhaustive 23-year follow-up with a much more captive audience in the Burghölzli Hospital, in which he reported course changes both for better and worse even after 20 years for a majority of his cases (Bleuler, 1978). Thus, “outcome” cannot be equated with Bleuler’s “end state.” No clear distinction was seen in the Kotov study between the outcome of schizoaffective disorder and schizophrenia, indicating that the DSM-IV/-5 diagnostic differentiation is not valid. Instead, co-morbidity between affective disorder and schizophrenia in the nonhierarchical DSM classification system is proposed.

The co-appearance of affective disorder and schizophrenia has always been acknowledged. Papa Bleuler included attacks of mania or melancholia in his list of etiopathogenetic “primary symptoms” (not to be confused with his symptomatological “basic disturbances”; see Bleuler, 1911). Kraepelin mentioned that episodes of mania and depression were not uncommon in schizophrenic patients and that quite a number of patients presented with symptoms that did not allow a confident distinction between manic-depressive insanity and dementia praecox (Kraepelin, 1920). He proposed as a plausible explanation that the presentation of symptoms was determined by predisposing factors in the patients’ personalities for emotional or schizophrenic manifestation of the manic-depressive or schizophrenic illness.

Odegaard, unconstrained by either DSM or ICD, and using the national Norwegian psychiatric register which he had tirelessly constructed, observed the diagnostic distribution of probands and (only) their psychotic relatives (Odegaard, 1972). He routinely saw affective psychoses in the relatives of schizophrenics, and schizophrenic psychoses in the relatives of atypical affective psychoses plus manic-depressive psychoses. He favored some kind of a polygenic theory for his results (compare to Gottesman and Shields, 1967).

Having prominent affective symptoms or syndromes in patients with schizophrenia eventually was considered to be a schizoaffective subtype of schizophrenia, and since DSM-III/III-R and –IV and ICD-10, schizoaffective disorder has been differentiated as an independent category; in DSM it is nearer to schizophrenia than in ICD because DSM requires at least two weeks of non-affective psychosis. The separate classification has been supported by validating genetic studies (Bertelsen and Gottesman, 1995; Hamshere et al., 2009) and a major register-based cohort study, indicating that schizoaffective disorder is genetically linked to both mood disorder and schizophrenia as an intermediate form (Laursen et al., 2005).

In a recent Danish register-based study of schizophrenia and bipolar disorder in offspring of two, one, or no parent likewise affected (Gottesman et al., 2010), we observed a cumulative incidence of bipolar disorder in offspring of two schizophrenic parents that was 10 times higher than in the general population, and of schizophrenia in offspring of two parents with bipolar disorder four times higher than the population value. In children of one schizophrenic parent and the other with bipolar disorder, the incidence of schizophrenia and of bipolar disorder was two to three times the incidence from only one parent affected with either disorder. A major Swedish population-based study provided similar evidence that schizophrenia and bipolar disorder share a common genetic cause (Lichtenstein et al., 2009). In a sophisticated, eclectic discussion of the not yet disappearing dichotomy, Craddock and Owen conclude that a broadly defined schizoaffective illness “may be particularly useful for genetic studies” (Craddock and Owen, 2010), reprising their earlier empirical results with the WTCCC cohort (Hamshere et al., 2009).

In order to get nearer to the relation to the genetic predisposition than the present classification allows, it has been suggested to study domains of symptoms, (the NIMH Research Domains Criteria project [RDoC]; see Insel et al., 2010), particularly in endophenotype studies (Insel and Cuthbert, 2009; Gottesman and Gould, 2003) as a promising way of future research of the basic relationships among the disorders behind what we, for the time being, term schizophrenia, schizoaffective disorder, and bipolar disorder. The earlier Research Diagnostic Criteria (RDC) of Spitzer et al. (Spitzer et al., 1978) and the OPCRIT of McGuffin et al. (McGuffin et al., 1991) anticipated less constrained approaches to diagnosis that have shown their merit in genetically promising research. We find the conclusions of Hamshere et al. (Hamshere et al., 2009) compatible with our current understanding: "We hope that psychiatry is moving towards the time when our patients can benefit from diagnostic concepts that are built on solid foundations of empirical biological evidence rather than being perched precariously on the shifting sands of expert opinion."


Kendler KS (2013) Psychosis Within vs. Outside of Major Mood Episodes: A Key Prognostic and Diagnostic Criterion. JAMA Psychiatry. Abstract

Maurano MT, Humbert R, Rynes E, Thurman RE, Haugen E, Wang H, Reynolds AP, Sandstrom R, Qu H, Brody J, Shafer A, Neri F, Lee K, Kutyavin T, Stehling-Sun S, Johnson AK, Canfield TK, Giste E, Diegel M, Bates D, Hansen RS, Neph S, Sabo PJ, Heimfeld S, Raubitschek A, Ziegler S, Cotsapas C, Sotoodehnia N, Glass I, Sunyaev SR, Kaul R, Stamatoyannopoulos JA. Systematic localization of common disease-associated variation in regulatory DNA. Science. 2012 Sep 7;337(6099):1190-5. Abstract

Cardno AG, Rijsdijk FV, West RM, Gottesman II, Craddock N, Murray RM, McGuffin P (2012) A twin study of schizoaffective-mania, schizoaffective-depression, and other psychotic syndromes. Am J Med Genet B Neuropsychiatr Genet. 159B(2):172-82. Abstract

Bertelsen A, Gottesman I I (1995) Schizoaffective Psychoses: Genetical Clues to Classification. American Journal of Medical Genetics (Neuropsychiatric Genetics) 60:7-11. Abstract

Laursen T M, Labourieau R, Licht R, Bertelsen A, Munk-Olsen T, Mortensen P B (2005) Family History of Psychiatric Illness as a Risk Factor for Schizoaffective Disorder. Arch Gen Psychiatry/vol 62: 841-848. Abstract

Gottesman I I, Laursen T M, Bertelsen A, Mortensen P B (2010) Severe Mental Disorders in Offspring with 2 Psychiatrically Ill Parents. Arch Gen Psychiatry vol 67(3): 252-257. Abstract

Lichtenstein P, Yip B H, Björk C, Pawitan Y, Cannon T D, Sullivan P F, Hultman C M (2009) Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: a population-based study. Lancet 373: 234-239. Abstract

Bleuler E (1911) Dementia præcox oder Gruppe der Schizophrenien. Deuticke Leipzig, Wien. English edition (1950) Dementia praecox or the group of schizophrenias. Intern. Univ. Press, New York.

Kraepelin E (1920) Die Erscheinungsformen des Irreseins. Z.f.d.g.Neur.u.Psych. LXII: 1-29.

Odegaard (1972) The multifactorial theory of inheritance in predisposition to schizophrenia. In: Kaplan, A.R., ed. Genetic Factors in "Schizophrenia." Springfield, III.: Charles C Thomas, Publisher, 1972. pp. 256-275.

Gottesman II, Shields J. (1967) A polygenic theory of schizophrenia. Proc Natl Acad Sci U S A. 1967 Jul;58(1):199-205. Abstract

Hamshere ML, Green EK, Jones IR, Jones L, Moskvina V, Kirov G, Grozeva D, Nikolov I, Vukcevic D, Caesar S, Gordon-Smith K, Fraser C, Russell E, Breen G, St Clair D, Collier DA, Young AH, Ferrier IN, Farmer A, McGuffin P; Wellcome Trust Case Control Consortium, Holmans PA, Owen MJ, O'Donovan MC, Craddock N. (2009) Genetic utility of broadly defined bipolar schizoaffective disorder as a diagnostic concept. Br J Psychiatry. Jul;195(1):23-9. Abstract

Craddock N, Owen MJ. (2010) The Kraepelinian dichotomy—going, going... but still not gone. Br J Psychiatry; 196(2):92-5. Abstract

Insel T R, Cuthbert B, Garvey M et al. (2010) Research domain criteria (RDoC): Toward a new classification framework for research on mental disorders. Am J Psychiatry. 167: 748-751. Abstract

Insel T R, Cuthbert B N (2009) Commentary: Endophenotypes: Bridging Genomic Complexity and Disorder Heterogenity. Biol Psychiatry, 66: 988-989. Abstract

Gottesman I I, Gould T D (2003) The Endophenotype Concept in Psychiatry: Etymology and Strategic Intensions. Am J Psychiatry 160: 636-645). Abstract

Spitzer RL, Endicott J, Robins E. (1978) Research diagnostic criteria: rationale and reliability. Arch Gen Psychiatry; 35(6):773-82. Abstract

McGuffin P, Farmer A, Harvey I. (1991) A polydiagnostic application of operational criteria in studies of psychotic illness. Development and reliability of the OPCRIT system. Arch Gen Psychiatry; 48(8):764-70. Abstract

Bleuler, M. (1978) The schizophrenic disorders: Long-term patient and family studies. Yale University Press.

View all comments by Irving Gottesman
View all comments by Aksel Bertelsen