Secondary Schizophrenia (75 page)

186

disease, with the exception of headache
[3].
The impact
tor in focusing the attention on this association. A
Chapter 13 – Schizophrenia-like psychosis and traumatic brain injury

systematic review of published case reports was per-based study and the ascertainment of schizophrenia
formed recently
[5],
which included 69 cases from 39

used a standard instrument (albeit one that has been
reports in the period 1971 to 1994. The authors applied
questioned in its validity), the ascertainment of head
DSM-IV criteria of “psychotic disorder due to trau-injury was crude and relied on self-report alone. The
matic brain injury (PDTBI)” with hallucinations or
cross-sectional nature of the study means that both
delusions being present, the psychosis being consid-schizophrenia and TBI are measured at the same time,
ered a direct consequence of the injury, and not being
and the nature of causality can be difficult to deter-accounted for by another mental disorder or delirium.

mine. It could be argued, for example, that individu-The crucial decision in such cases is meeting the “direct
als predisposed to schizophrenia are behaviorally more
consequence” criterion. The usual approach taken by
predisposed to a head injury, an instance of reverse
clinicians is that there is a temporal association (TBI
causality, which this design cannot clarify.

preceding SLP, generally by a short interval but with
no upper time limit), the individual was free of psy-Case control studies
chopathology, in particular, prepsychotic symptoms
prior to the TBI, and there was no apparent genetic
These studies have been used to both examine the
vulnerability to the development of schizophrenia (i.e.

strength of the association and identify risk factors for
no family history of psychosis). They might also take
the development of SLP following TBI.

the type of injury into consideration. A closed-head
injury with concussion and presumed fronto-temporal
Studies examining the strength

injury, and an open injury with trauma to the frontal
or temporal lobes may be considered as “sufficient
of the association

evidence.”

Evidence has been presented
[7, 8]
to implicate preg-On the other hand, an injury exclusively to the
nancy and delivery complications, some of which
parietal or occipital lobe is considered as insufficient
result in TBI, as an increased risk for the development
evidence for such an association, based on the func-of schizophrenia. In a large case-control study, Wilcox
tional neuroanatomy of schizophrenia. Neuropsycho-and Nasrallah
[9]
used medical records of patients
logical deficits may also be considered in the deter-admitted to a university hospital between 1934 and
mination, with fronto-subcortical or temporal lobe
1944 with schizophrenia (n
=
200), depression (n
=

deficits suggesting plausibility. Unfortunately, because
203), mania (n
=
122) and “surgical” controls (n
=

the anatomical and neuropsychological deficits in
134) to examine the prevalence of head injury before
schizophrenia are not specific, they do not offer clear
age 10 years. The odds ratio (OR) for schizophrenia
guidelines for meeting the “direct consequence” crite-for those with childhood head injury was 16.6 (CI 2.6

rion, and clinical judgment is generally relied upon.

to 689), a highly significant result. The odds for bipo-In conclusion, anecdotal reports of SLP following
lar disorder (OR 6.9, CI 0.8 to 321) and depression
TBI do not and cannot provide strong evidence for the
(OR 2.0, CI 0.2 to 10.5) were nonsignificantly elevated.

association because of the likelihood of chance asso-These authors did not find any significant relationship
ciation and the subjective nature of the determination.

with a particular location of the injury. Gureje and col-However, they have prompted more systematic studies.

leagues
[10],
in a case-control study in a Nigerian sample, also found an association between childhood brain
trauma and schizophrenia. These patients also had
Cross-sectional surveys

mixed laterality in adulthood, possibly attributable to
The noteworthy cross-sectional study was conducted
left hemispheric damage.

as part of the U.S. National Epidemiological Catch-The major limitation of studies that rely on histor-ment Area study based in New Haven, Connecticut
[6],

ical information is recall bias, with the suspicion that
with any heads injury leading to loss of consciousness
subjects with schizophrenia and their family members
or confusion being included. Of the 361 (7.2%) head-are more likely to report past TBI than matched con-injured individuals in the sample, 3.4% were diagnosed
trols. There is also a likelihood of observer bias, with a
with schizophrenia (cf. 1.9% of controls without head
clinician more likely to inquire and record an episode
injury), a statistically nonsignificant result (chi2
=
2.8,
of head injury in a neuropsychiatric patient than a con-

187

p
=
0.093). Although this was a large population-trol patient. The issue of a suitable control population
Organic Syndromes of Schizophrenia – Section 3

is another consideration. These limitations have been
ing identified a large birth cohort (n
=
785,051), in
overcome by two case register linkage studies from
whom 748 cases of schizophrenia and 14,960 matched
countries in which population-based case registers are
controls were identified and a documented history of
maintained and most of medical encounters occur in
head injury obtained through linkage with the vari-the public sector.

ous health registers in Sweden, which is recognized
The study by Nielsen and colleagues
[11],
a nested
for the comprehensive nature of such documentation.

case-control study, identified 8,288 individuals with
This study found no increase in schizophrenia after
ICD-8 schizophrenia who had been admitted to a Dan-head injury (OR 1.10, CI 0.82 to 1.47), and a small
ish psychiatric hospital between 1978 and 1993, and
increase of nonschizophrenic, nonaffective psychosis
matched each case to 10 controls from the general pop-

(OR 1.37, CI 1.14 to 1.66), which occurred only in
ulation, based on the Central Persons Register. The
head injuries during adolescence or later. The asso-subjects were linked to the National Patient Register
ciation was not appreciably influenced by the inclu-to identify any admission secondary to TBI between
sion of various confounding factors (year of birth,
1978 and the date of the index psychiatric admission.

highest parental income, highest parental education,
The authors also noted the occurrence of fractures
highest parental occupation, area of birth, family his-affecting other parts of the body in this period, as
tory of psychosis, Apgar score at 5 min, gestational
an index of proneness to accidents and injury. Over-age, paternal age, birth weight, birth length) in the
all, there was no excess of concussion or severe head
model.

injury in the schizophrenia patients (OR 0.94 and 0.89,
This is the largest case-control study in this field
respectively; not a significant result). Individuals with
and its reliance on record linkage and a birth cohort
schizophrenia were less likely to have had another frac-gives it much strength, although it suffers from the
ture (ORT
=
0.71, p < 0.01), and if this was corrected
same limitation as the Danish study in terms of the
for, head injury was slightly but significantly more
diagnoses being made in different centers with lack
likely to be associated with schizophrenia (OR 1.37

of standardization, because organic psychoses were
and 1.29, respectively, p < 0.01). The interpretation of
not included. Overall, this excellent study, in com-fewer fractures in schizophrenic patients is problem-bination with the previous literature, makes it very
atic, appearing to suggest that in the prodromal phase
unlikely that there is a strong link between head injury
of schizophrenia, their behavior such as social with-and schizophrenia, but a weak association is not ruled
drawal, made them less likely to suffer fractures (a case
out.

of reverse causality). This was complicated by the fact
Two other studies used the case-control method
the rates of fractures differed depending upon the lag
to ask the question whether TBI increased the risk of
periods between injury and psychosis. The authors also
schizophrenia in individuals at genetically increased
examined the risk in relation to the interval between
risk of the disorder. Corcoran and Malaspina
[13]
used
TBI and schizophrenia, and reported that head injury
the diagnostic interview for genetic studies to inquire
was more likely in the schizophrenia group in the 12

about a history of significant head injury in individu-months preceding the index admission (OR 2.0 and
als from multiplex schizophrenia pedigrees (n
=
561)
1.84, respectively, p < 0.01) and less likely beyond the
and multiplex bipolar disorder pedigrees (n
=
1,271)
1-year interval.

participating in a genetics study. Overall, the individ-Although this study should be considered to be
uals with schizophrenia were > 3 times more likely to
largely negative, it does not rule out a small but signif-report a previous TBI than unaffected individuals. This
icant increase in the risk of schizophrenia in the year
risk increased with the genetic risk of schizophrenia. In
following TBI. The study, through the linkage of reg-the bipolar pedigree, there was no increased risk, and
isters, overcame recall bias. However, observer bias is
it was fourfold in those with a high genetic risk for
not ruled out, as it is possible that the history of head
schizophrenia (OR 4.27, CI 1.40 to 13.0). The authors
injury prompted clinicians to make a diagnosis of an
did not test for an interaction effect, and we therefore
“organic” psychosis rather than schizophrenia, leading
are not certain whether the increased risk is specific to
to an underestimate of the association; the retrospec-the high genetic risk group.

tive design does not pick up this detail.

A Canadian study
[14]
also used the case-The recent nested case-control study by Harrison
control design in 23 multiply-affected families with
188

and colleagues
[12]
has the added advantage of hav-schizophrenia and reported that schizophrenia
Chapter 13 – Schizophrenia-like psychosis and traumatic brain injury

patients had an excess of TBI sustained before 10

authors concluded that the risk in the brain-injured
years (OR 2.34, CI 1.03 to 5.03) or before 17 years (OR

group was two-to threefold. The limitations of hetero-1.90, CI 0.95 to 3.79), the former being statistically
geneity of diagnostic criteria applied, and the use of a
significant.

population based incidence figure derived by different
In summary, the case-control studies provide
methodology, are obvious.

strong evidence that TBI does not substantially
The most well-known cohort study was published
increase the risk of schizophrenia in the general popu-from Finland
[18]
and was a 22–26 year follow-up
lation, although a small increase cannot be ruled out.

of 3,552 Finnish veterans of World War II. Almost
In individuals with a genetic risk for schizophrenia,
all (98.8%) suffered injuries from bullets or shrapnel
this increase in risk may be greater, although the ret-and 42% of injuries were open, making them atypi-rospective nature of the studies make the data prone to
cal of peacetime injuries. Medical care for the veter-bias.

ans was provided by one hospital, and the records of
this hospital were examined for history of psychosis.

Studies that examined risk factors for

Defined broadly, 317 (8.9%) individuals had the onset
of psychosis following brain injury; of these, 2.1%

schizophrenia following TBI

had schizophrenic psychoses and another 2.0% para-The study by Sachdev and colleagues
[15]
included 45

noid psychoses, giving an incidence of SLP of about
patients with SLP following TBI who had been referred
4.1% (CI 3.5 to 4.7%). A significant proportion (42%)
for a medicolegal opinion and were so determined by
had the onset of psychosis more than 10 years after
the authors, and matched them with 45 head-injured
the injury. Although the study strongly supported an
patients without SLP or other psychiatric disorder. The
increased incidence of SLP, it suffers from the lim-authors found that a family history of schizophrenia
itations of observer bias, lack of standardized diag-was a risk factor for SLP, but the nature and degree
nostic assessments, the atypical nature of brain injury,
of head injury did not produce a significant effect,
the possibility of psychological factors such as combat
although the study was not powered for subtle effects.