Professional Documents
Culture Documents
Background: Numerous epidemiological studies found Results: A number of specific risk factors were associ-
an increased risk of schizophrenia among persons ex- ated with an increased risk of schizophrenia. The rela-
posed to various obstetric complications. The underly- tive risk (95% confidence interval) for preeclampsia was
ing mechansims are unknown. 2.5 (1.4-4.5); vacuum extraction, 1.7 (1.1-2.6); and mal-
formations, 2.4 (1.2-5.1). In logistic regression models,
Objective: To study specific risk factors, as well as sets we found that indicators of all 3 etiologic mechanisms
of risk factors, representing 3 different etiologic mecha- were associated with increased point estimates of schizo-
nisms: (1) malnutrition during fetal life; (2) extreme pre- phrenia, although at lower risk levels. Preeclampsia, an
maturity; and (3) hypoxia or ischemia. indicator of fetal malnutrition, was the only risk factor
with statistically significant increased risk after control
Methods: In this longitudinal cohort study, informa- for all potentially confounding factors.
tion in the National Birth Register was linked to the Na-
tional Inpatient Register. We followed up 507 516 chil- Conclusion: This study supports the theory of an asso-
dren born between 1973 and 1977 with regard to a ciation between obstetric complications and schizophre-
diagnosis of schizophrenia between 1987 and 1995 (238 nia. Although preeclampsia was the strongest indi-
cases). By record linkage, we also had access to data on vidual risk factor, there was evidence of increased risk
psychiatric illness in the mother. Occurrence of schizo- associated with all 3 etiologic mechanisms.
phrenia was measured by the Mantel-Haenszel test and
logistic regression. Arch Gen Psychiatry. 1999;56:234-240
T
HE ORIGIN of schizophre- gives little clue to the underlying cause. Mc-
nia is largely unknown. A Neil et al6 showed that the kind of scale used
vulnerability originating has consequences for the results obtained,
from disturbances in the with higher risk estimates if a weighted scale
central nervous system is used.7
(CNS) during the embryonal, fetal, or Several studies found 1 or 2 (rarely
neonatal period may possibly explain 3) significant risk factors that are seldom
some of the reason for the occurrence of confirmed in other studies. Suggested risk
schizophrenia.1 factors are preeclampsia,8-10 small head cir-
From the Stockholm County Numerous epidemiological studies cumference,11,12 low birth weight,13-17 Rh
Child and Adolescent Public
Health Unit (Dr Dalman) and
have been published on the association be- incompatibility,18 fetal distress,19 body
the Department of Obstetrics tween obstetric complications and schizo- weight heavy for length (weight/length, $1
and Gynecology phrenia. In a meta-analysis2 of 18 stud- SD),20 and abnormal presentations.21-23
(Dr Grunewald), Huddinge ies, an overall odds ratio of 2.03 (95% Most of these studies are relatively small,
University Hospital; the Center confidence interval [CI], 1.6-2.4) was including fewer than 100 patients, and,
for Psychosis Research, South found for schizophrenia following obstet- thus, the results are not precise regard-
Stockholm Health Authority ric complications of any kind. Interest- ing risk estimates. In some studies,13,24 ma-
(Dr Cullberg); and the Center ingly, the only 2 historical cohort stud- ternal recall has been used instead of case
for Epidemiology, National ies 3,4 in that analysis failed to find a records, which implies an information
Board of Health and Welfare
significant association. However, the num- bias risk.
(Mr Köster), Stockholm,
Sweden; and the Department of ber of schizophrenic or psychotic pa- Other types of evidence for prenatal
Social Medicine, Vasa Hospital, tients was small in those studies. risk factors for schizophrenia are the in-
University of Gothenburg Many studies have used the sum- creased occurrence of minor physical
(Dr Allebeck), Gothenburg, mary scale of obstetric complications by anomalies (for example, epicanthal fold
Sweden. Lewis et al,5 which is unspecific and thus and syndactylia) in individuals with
By using the National Birth Register,41,42 we obtained data Psychiatric Illness in Mothers
on obstetric complications in all Swedish children born be-
tween 1973 and 1977 (507 516 children). The National Birth Data on the mother in the birth records enabled us to search
Register has been in operation since 1973 and contains in- for data on psychiatric care in the National Inpatient Reg-
formation on all children born in Sweden, around 100 000 ister from 1973 onward. We collected data on psychotic
every year. Data are based on forms completed by mid- illness (codes 295-299 according to ICD-8 until 1986; codes
wives and the pediatricians in charge and sent to the Na- 295-298 according to ICD-9 from 1987) in the mothers for
tional Board of Health and Welfare with information on risk the period from 1973 to 1995.
factors in pregnancy, events during delivery, anthropomet-
ric data on the child, and conditions present in the neona- CLASSIFICATION OF ETIOLOGIC MECHANISMS
tal period. Diagnoses were given according to Interna-
tional Classification of Diseases, Eighth Revision (ICD-8).43 Variables reflecting the 3 types of etiologic mechanisms de-
The proportion of missing data is around 1%. scribed herein were grouped to identify states reflecting the
Anthropometric data and malformations are classified following:
in the register. Small for gestational age and large for gesta- 1. Fetal malnutrition: preeclampsia during preg-
tional age are classified according to a method used by the nancy (ICD-8 code 637.03-637.99), small for gestational
National Board of Health and Welfare based on birth weight age, ponderal index less than 0.2, which is a measurement
in relation to gestational age (based on information regard- of leanness (weight/length3).
ing the last menstruation). Cutoff points for these and other 2. Extreme prematurity: delivery before the onset of
anthropometric measures are defined according to Maršál week 33 (based on information about last menstruation).
et al44 (standard curves ± 2 SDs based on mean measure- 3. Hypoxia or ischemia around birth: cesarean sec-
ments of 759 ultrasonic estimates of Scandinavian chil- tion and vacuum extraction in combination with a diag-
dren). Malformations are classified into types 1 and 2 in the nosis of threatening fetal distress or intrauterine anoxia
register. Type 1 includes all registered malformations ex- (ICD-8 codes 776.30-776.40), breech delivery (ICD-8 codes
cept preauricular appendix, retractile testis, hydrocele tes- 650.6-662.6), placental abruption (ICD-8 code 651.4), an
tis, hip-joint luxation, and nevus. Type 2 includes malfor- Apgar score of 0 to 6 at 1 and 5 minutes, respectively, ac-
mations that are known to be more accurately and completely cording to a definition of neonatal distress.46
recorded than those in type 1. These are malformations in
the CNS, lungs, and intestinal canal, serious malformations DATA ANALYSIS
of the heart, ears, or eyes, choanal atresia, facial clefts, esoph-
ageal atresia, intersexuality, and hypospadias. We calculated relative risk (RR) estimates with the Mantel-
Haenszel test and 95% CIs.47 Logistic regression analysis
PSYCHIATRIC RECORDS was used to take into account other obstetric complica-
tions as confounders for each other. By logistic regression
Psychiatric Illness in Offspring analyses, we estimated odds ratios (ORs) for particular vari-
ables, controlling for the other variables included in each
By means of the unique personal identity number, the co- hypothesized etiologic mechanism. We also used logistic
hort was followed up in the National Inpatient Register that regression analysis to study the impact of each mecha-
covers between 97% and 98% of all episodes of inpatient nism while controlling for the other 2 hypothesized mecha-
care. They were given clinical diagnoses according to the nisms. Finally, we constructed a logistic regression model
Nordic version of International Classification of Diseases, with the most important variables from all 3 mechanisms.
schizophrenia25,26 and findings from the Dutch Famine plain the association between obstetric complications and
Study by Susser et al,27 in which malnutrition was found schizophrenia. In particular, we wanted to distinguish these
to be related to neurodevelopmental disturbances in- mechanisms since they require different preventive ap-
cluding schizophrenia. proaches and also correlate, with an overlap, to different
Men have an earlier onset of illness, lower premor- types of neuropathologic characteristics.34
bid functional level, and poorer outcome than wom- 1. Reduction in the supply of nutrients, such as oxy-
en.28,29 Thus, an interesting question is whether there is gen, iodine, glucose, and iron, to the fetus may lead to
a sex difference in the risk of schizophrenia following ob- impaired development of the CNS35 as well as intrauter-
stetric complications; the results of previous studies are ine growth restriction. In these cases, the lack of me-
inconsistent in this respect.11,30-33 Another question that tabolites and states of hypoxia are repeated over time and
has been raised is whether obstetric complications are the basal ganglia is at particular risk of being dam-
more common in cases with an early age of onset.8,22,30 aged.34 A more narrow definition of this state would be
Central nervous system disturbances can arise through chronic fetal hypoxia.4
several mechanisms. In this study, we considered the fol- 2. Prematurity increases the risk for intracranial
lowing 3 possible etiologic mechanisms that might ex- hemorrhages,36 periventricular leukomalacia,34 and also
for interstitial respiratory distress syndrome and infec- (RR, 2.3 [95% CI, 1.1-4.7]). The risk was also in-
tions,37 which may also cause brain damage. creased, although it was not statistically significant, if the
3. Hypoxia or ischemia due to complications dur- mother was unmarried or if her marital status was un-
ing delivery could result in brain damage, especially in known at the time of delivery.
the regions of the hippocampus and cortex.34 Table 2 shows the RRs of schizophrenia among chil-
Although all these etiologic mechanisms are plau- dren born with various obstetric complications. We found
sible, there is also an association between a psychotic dis- that several risk factors during pregnancy, as well as dur-
order in the mother and an increased risk of obstetric com- ing the time around delivery, significantly increased the
plications (shown by Sacker et al38 and also in our data). risk of schizophrenia: preeclampsia, gestational age
Therefore, the occurrence of psychotic disorders in moth- younger than 33 weeks, inertia of labor, vacuum extrac-
ers should be taken into account. tion, a ponderal index less than 20, respiratory illness,
The possibility of linking the Swedish National and type 2 malformations. Adjustments for the vari-
Birth Register with the National Inpatient Register39,40 ables presented in Table 1 did not dramatically alter the
enabled us to address several of the issues mentioned results. There was a clear difference between the sexes.
herein. Data on obstetric complications were available Being the fourth child and small for gestational age were
for all Swedish children born between 1973 and 1977 associated with a significantly increased risk of schizo-
(507 516 children). The children were observed with phrenia in males, while a birth weight of less than 1500
regard to diagnosis of schizophrenia to a maximum g significantly increased the risk of schizophrenia among
age of 22 years. Furthermore, the birth records could females.
be linked to data on psychiatric illness in the mother, In Table 3, we analyzed by logistic regression the
which enabled us to control for maternal psychotic impact of the variables included in each etiologic group.
illness in the analyses. We wanted to study specific Among the hypoxia or ischemia variables, low Apgar
risk factors as well as sets of risk factors that were scores at 5 minutes and vacuum extraction or cesarean
indicators of the 3 etiologic mechanisms mentioned section on indication of fetal distress were associated with
herein. the highest risks, although these risks were not statisti-
cally significant. Among the indicators of malnutrition,
RESULTS preeclampsia and low ponderal index (leanness) were the
strongest risk factors. The third group consisted only of
The RRs associated with the major risk factors are pre- 1 factor, extreme prematurity, which was associated with
sented in Table 1. These factors are controlled for in a significantly increased risk. The separate ORs for each
the analyses presented below. There was an 8-fold in- mechanism, without adjustments for the 2 other mecha-
creased risk of schizophrenia among offspring to moth- nisms, are also shown in Table 3.
ers who had had a psychotic disorder (including schizo- In the next set of regression models, we studied the
phrenia) during their adult life. The risk of obstetric impact of each etiologic mechanism adjusted for the other
complications, defined as any of the complications in- 2 mechanisms. The results are shown at the bottom line
cluded as indicators of the 3 mechanisms, was also in- of the specific mechanisms in Table 3. There is an in-
creased (RR, 1.3 [95% CI, 1.2-1.3]) in this group of moth- creased risk, although the risk was not statistically sig-
ers. Children born to mothers younger than 19 or older nificant, among those who had been exposed to indica-
than 39 years had an increased risk of schizophrenia, al- tors of fetal malnutrition (a total of 24 individuals with
though this risk was not statistically significant, except schizophrenia) and a gestational age younger than 33
for males born to mothers older than 39 years (RR, 3.5 weeks (5 individuals). In the hypoxia or ischemia group,
[95% CI, 1.4-8.9]) and females born to young mothers the risk estimate was lower.
*Controlling for confounders listed in Table 1. Ellipses indicate not applicable. See “Materials and Methods” for definition of type 1 and type 2 malformations.
†Adjusted for psychotic illness in mother, mother’s age and marital status, the child’s year of birth, birth hospital, and sex.
‡Adjusted for psychotic illness in mother, mother’s age and marital status, the child’s year of birth, and birth hospital.
Table 4. Odds Ratios (ORs) for Schizophrenia: Comparison Between the Strongest Risk Factors in a Logistic Regression Model
*Adjusted for psychotic illness in mother, mother’s age and marital status, the child’s year of birth, and birth hospital.
†Adjusted for psychotic illness in mother, mother’s age and marital status, the child’s year of birth, birth hospital, and sex.
By comparing the unadjusted and the adjusted ORs maturity (OR, 4.3 [95% CI, 1.0-18.1]). Proportionally
for each of the 3 mechanisms in Table 3, we were able to more males were exposed to the obstetric complications
study the effects of confounding between the different (Table 3).
mechanisms. The risk estimate for prematurity was re- In a final model, the strongest risk factors from each
duced, which could be interpreted as if some of the risk etiologic group in Table 3 were entered in a logistic re-
was explained by factors of malnutrition during preg- gression model to assess the separate effects of the most
nancy and hypoxia or ischemia at birth. The ORs of mal- important risk factors of the 3 etiologic mechanisms
nutrition and hypoxia or ischemia were less reduced in (Table 4). The risk associated with a low Apgar score,
this procedure. Notably, the risks were still increased, low ponderal index, and prematurity was reduced, es-
which indicates independent effects. pecially among the females, compared with the findings
The risks were unevenly distributed between the in Table 3. However, the point estimates were increased
sexes (Table 3). Indicators of malnutrition were associ- for complications of all 3 mechanisms, although only pre-
ated with higher risk in males (OR, 2.1 [95% CI, 1.3- eclampsia remained significantly associated with an in-
3.4]), while the females had increased risk following pre- creased risk of schizophrenia.