In 2004, Murray et al. (1) suggested that the main difference between bipolar disorder (BPD) and schizophrenia (Sz) was a larger prevalence of developmental impairments in the latter. A number of recent articles in Acta Psychiatrica Scandinavica suggest that categories of affective and non-affective psychosis may be similar in their pattern of associations with measures as diverse as social cognition (2), brain volumes (3) and metabolic dysregulation (4), begging the question what the actual status is of the prediction offered by Murray et al. more than 5 years ago. These issues are particularly important given the upcoming revisions of diagnostic systems in psychiatry: how should bipolar disorder and schizophrenia be classified in DSM-V and ICD-11?

A review was conducted of studies focussing on this issue, with a special focus on work published in the last 10 years; results are displayed in Table 1 below.

Table 1.   Review of critical contrasts and similarities in schizophrenia and bipolar disorder
Type of variableVariableSchizophrenia (Sz)Bipolar disorder (BPD)Comment
Psychometric risk states in general populationSubclinical psychometric risk states in representative population samplesSubclinical psychosis prevalence around 5–10% (6)Subclinical bipolarity prevalence around 10–20% (7)Strong comorbidity between psychosis and bipolar psychometric risk states in population (8), similar to the comorbidity seen in clinical disorders (9)
Transition from subclinical psychometric risk states to clinical disorder in epidemiological studiesAt a rate of around 4% over 1 year follow-up (10)At a rate of around 3% over 1 year (11)
PsychopathologyPositive symptomsPresentPresentSz>BPD, diagnostic contrast low(12)
Negative symptomsPresentPresentSz>BPD, diagnostic contrast low(12)
DisorganisationPresentPresentSz>BPD, diagnostic contrast low(12)
Mania symptomsPresentPresentBPD>Sz, diagnostic contrast high(12, 13)
Depression symptomsPresentPresentBPD>Sz, diagnostic contrast high(12, 13)
EpidemiologyLifetime prevalence≈1%≈1%High P:I ratio indicates high proportion chronic course; Onset and sex differences shifted to later ages in BPD compared to Sz
Prevalence:incidence ratio≈100≈100
Age of onsetDevelopmental pattern(14)Developmental pattern(15)
Sex differences onsetEarlier in men(16)Earlier in men(17)
Risk factorsCannabisPositive association(18)Positive association(19)Generally much fewer studies in BPD; meta-analyses largely in Sz only.
Urban upbringingPositive association(20)No association(21, 22)
Ethnic groupPositive association(23)Positive association(24, 25)
Parental agePositive association(26)Positive association(27, 28)
Prenatal faminePositive association (early)(29)Positive association (late) (30)
Pregnancy complicationsReviews suggest association with Sz(31) but not BPD(32). However, reviews largely based on retrospective studies; suggestion of increased risk for BPD in rare prospective studies(30, 33, 34)
GeneticsRelative with BPDPositive association(35)Positive association(35)Molecular genetic studies (GWAS) suggest genetic variation common to Sz and BPD e.g. ZNF804A and CACNA1C(36)
Relative with SzPositive association(35)Positive association(35)
CognitionNeurocognitionImpaired(37, 38)Impaired(37, 38)Sz>BPD
Mentalizing abilityImpaired(2)Impaired(2)Sz>BPD; may be mediated by neurocognitive alterations(2, 39)
Intermediary phenotypesNeurocognitive impairmentGeneralised alterations associated with familial/genetic risk for Sz(40, 41)Very few and specific alterations associated with familial risk for BPD(37, 42)Other measures, e.g. neurophysiology, too few comparative data, some reports suggest similarities(43)
Developmental alterationsChild cognitive abilityAlterations (44–46)No alterations(47–49) 
Child motor developmentAlterations(44–46)No alterations(47–49)
Child emotional developmentAlterations(44–46)Alterations(47–49)
Neuroradiological findings and other biological markersMany individual studies but to date no systematic quantitative integration of findings of Sz vs. BPD. Uncertain whether this will ever be possible given heterogeneity of measures, methodologies, patient selection, unmeasured confounding and differential neurotrophic effects of medications(50). Fronto-temporal systems may be most affected in both Sz and BPD. Ventricular enlargement, cortical gray matter reductions and abnormal gyral folding (51) reported in both Sz and BPD, data suggest Sz>BPD; suggestion of qualitative differences in hippocampal and amygdalar volumes in Sz and BPD.

Both Sz and BPD have onset in adolescence and young adulthood, with an earlier onset in men. The incidence of both disorders is low and associated with a high prevalence:incidence ratio, indicative of a high rate of chronicity. Schizophrenia patients have high rates of affective symptoms, although lower than in BPD, and bipolar patients have negative symptoms and cognitive impairment, although lower than in Sz. The diagnostic contrast between Sz and BPD is high for affective symptoms, but this is mostly due to artificial diagnostic exclusion criteria.

Sz and BPD have correlated genetic liabilities, some of which is becoming substantiated in recent genome-wide molecular genetic association studies. Of interest, however, is the fact that genetic risk for Sz is strongly expressed as neurocognitive impairment whereas genetic risk for BPD is only weakly expressed in the neurocognitive domain. Danish studies have shown that BPD and Sz have a high population comorbidity index; general population studies have shown that both BPD and Sz phenotypes are associated with psychometric risk states in healthy individuals, and that BPD and Sz psychometric risk states are similarly highly comorbid with each other. Risk factors representing social stress and defeat as well as early emotional alterations are associated with both BPD and Sz whereas risk factors reflecting early motor and cognitive alterations appear specific for schizophrenia. Interesting is the differential association with growing up in an urban environment (not associated with bipolar disorder), suggesting that this exposure impacts on specific developmental alterations associated with schizophrenia.

The literature on neuroimaging, biological variables and prenatal life suggests differences and similarities as regards their prevalences in Sz and BPD; firm conclusions, however, cannot be drawn because of many methodological constraints and the lack of direct comparisons.

In conclusion, science is catching up with century-old diagnostic traditions in psychiatry. It is becoming apparent that due to the strict separation imposed by DSM and ICD on the domains of affective and non-affective psychotic disorder, major opportunities have been missed to study the causes and treatment of psychiatric disorders. Psychotic disorders appear to be originating from (partly) overlapping areas of risk, one more developmental associated with cognitive impairment, and one more associated with affective dysregulation. It has been suggested that where these areas of risk overlap and interact with environmental risks, delusions and hallucinations may ensue through a final common pathway of dopamine dysregulation and affective cognitive biases (5). Therefore an elegant way of adapting the revisions of DSM and ICD diagnostic manuals to the scientific reality is to keep diagnostic categories more or less as they were and add cross-disorder dimensions of psychopathology, as depicted in Fig. 1. The simple addition of dimensions would finally clear the way for an official cross-diagnostic approach in psychiatry, ending a dysfunctional and rigid system of artificial partitioning.


Figure 1.  A combined dimensional and categorical system of diagnosis of psychotic disorders. Categorical diagnoses of schizophrenia (yellow), bipolar disorder (brown), and schizoaffective disorder (pink) are accompanied by a patient’s quantitative scores (connected by red lines) on five main dimensions of psychopathology.

Download figure to PowerPoint


  1. Top of page
  2. Acknowledgement
  3. References

Nil Kaymaz is supported by the Netherlands Organisation for Scientific Research (NWO) under project number: 017.002.048.


  1. Top of page
  2. Acknowledgement
  3. References