International Journal of Epidemiology Vol. 21, No.

2
© International Epidemiological Association 1992 Printed in Great Britain

Perinatal TORCH' Infections

Identified by Serology: Correlation
with Abnormalities in the Children
through 7 Years of Age
JOHN L SEVER, JONAS H ELLENBERG. ANITA C LEY. DAVID L MADDEN. DAVID A FUCCILLO,
NANCY R TZAN AND DOROTHY M EDMONDS

Sever J L (National Institute of Neurological Disorders and Stroke, National Institute of Health, Bethesda, Maryland
20892, USA), Ellenberg J H, Ley A C, Madden D L, Fuccillo D A, Tzan N R and Edmonds D M. Perinatal TORCH'
infections identified by serology: correlation with abnormalities in the children through 7 years of age. International

Downloaded from ije.oxfordjournals.org by guest on January 6, 2011

Journal of Epidemiology 1992; 21: 285-292.
A matched case-control methodology was used to assess the risk for a wide range of abnormalities in children
associated with serological evidence for 'TORCH' infections in the mothers. Specimens were selected from the large
bank of sera from the approximately 54000 pregnant women who participated in the Collaborative Perinatal Project.
There was no clear association between any of the antigens studied and any specific damage to the child. These
'negative' findings are consistent with the absence of frequent significant effects due to these agents in the second and
third trimesters of pregnancy.

The association between maternal infections with
Toxoplasma gondii, rubella, cytomegalovirus, and
Herpes simplex, the 'TORCH' infections, during preg-
nancy and certain types and frequencies of damage
to the children of these pregnancies has been well
established.1 Most studies have concentrated on one
infection at a time and the length of paediatric obser-
vation has been variable.2 A few studies have used pro-
spective serological testing to identify the maternal
infections.3
The present investigation employed a matched case-
control methodology to assess the risk for a wide range
of abnormalities in children associated with serological
evidence of 'TORCH' infections in the mothers. We
took advantage of the large bank of serial serum
specimens from approximately 54000 pregnant women
who participated in the Collaborative Perinatal Pro-
ject of the National Institute of Neurological and Com-
municative Disorders and Stroke (NCPP) and the data
from the 7-year follow up of the children. In the
NCPP the pregnant women were studied prospectively
from their first prenatal visit through delivery to 6
National Institute of Neurological Disorders and Stroke National
Institutes of Health, Belhesda, Maryland 20892. USA
Reprint requests to: John L Sever, Children's National Medical Center
111 Michigan Ave., NW Suite 2108 Washington. DC 20010, USA
weeks postpartum and serial blood samples were taken
throughout their pregnancy and postpartum. Women
in the NCPP whose children had certain abnormalities
(e.g. fetal death, congenital heart disease, etc.) were
matched with control mothers who did not have these
abnormal outcomes and the sera were tested under
code for antibodies to the 'TORCH' agents. The fre-
quencies of seroconversions and significant increases
in antibodies for the abnormals and controls were
compared.
METHODS
About 54000 pregnant women entered the NCPP bet-
ween 1959 and 1966, when they registered for prenatal
care at one of the 12 co-operating university hospitals.
The general methodology and sampling frame for this
study has been described elsewhere.4 Of this popula-
tion 46% was white, 46% was black, and most of
the remainder were of Hispanic origin. Overall, the
population was slightly lower in socioeconomic status
than the US population.5 The pregnant women were
scheduled for serial serum specimens at the time of
registration, every 2 months throughout pregnancy, at
delivery, and 6 weeks postpartum. Serum specimens
were aliquoted, frozen, shipped to the National In-
stitutes of Health and held at the Serum Center of the
Infectious Diseases Branch, NINDS for later study.

285
286 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Children born into the project underwent a series
of examinations, among which were standardized
physical and neurological examinations at newborn,
1 year, and 7 years, a psychological examination at 4
years of age, and a speech, language and hearing
examination at 3 years. Standardized follow-up of the
children continued through 1973, when the last child
completed the 7-year follow-up examination.
Seventeen abnormalities were selected for study, and
are listed in Table 1. A case/control methodology was
employed to examine the association of infection with
each abnormality. This approach required serological
testing of only a fraction of the entire cohort of over
54000 women, for the five antigens selected for
study. Matching was introduced to control for con-
founding factors and thereby increase the validity of
the inferences.
the unweighted sum of the squares of the differences
between each of the factors, LMP (in weeks), age (in
years), date of earliest serum specimen and date of
latest serum specimen (in weeks) was computed. Three
such controls were selected, ranked according to their
closeness to the cases. The closest control was chosen
unless insufficient serum was available for testing, in
which case the next closest subject was chosen. This
procedure gave least importance to the age difference
criterion. Controls were used only once.
Before selection of matched controls, both cases and
controls were screened for the adequacy of their blood
samples. All cases and controls were required to have
both an early (<28 weeks gestation) and a late blood
sample (37 weeks gestation to 12 weeks post-delivery).
For the cases, the earliest and latest bloods were used;
for the controls, the bloods closest to the abnormals
were used. During the first phase of the study, defined

Downloaded from ije.oxfordjournals.org by guest on January 6, 2011

TABLE 1 Abnormal outcomes in case-control studies only by the time frame in which the serological testing
was accomplished, the early and late bloods for both
Phase of study cases and controls were required to be at least 90 days
Number of cases in which sero- and no more than 300 days apart. During the second
Abnormal with adequate logical testing
outcomes blood samples was performed
phase, the minimum separation restriction was relaxed
to 60 days (with the exception of fetal deaths, abor-
Fetal death 580 11
tions or stillbirths). This resulted in an increase in the
Birthweight ^2000 g 577 1 proportion of adequate blood samples in the first
Neonatal seizures 122 . II phase of 56% to the second phase of 82%. With the
Neonatal death 296 1 exception of the outcome of low birthweight, the pro-
Congenital hearl disease 258 I portion of cases without adequate blood samples in
Congenital malformations 2769 I
Newborn bilirubin > I 7 mg^o 569 1
Phase I was fairly consistent for each of the abnor-
Microcephaly 599 II malities reported. This does not suggest a bias relating
Macrocephaly 373 II to the presence of adequate serum to the severity of
Hypotonia 100 1 morbidity in the child.
Delayed motor development 371 1
Fine or gross motor development 402 1 The few cases with no matched controls were
Abnormality of liver at 1 year 249 1 excluded from the study as were the few mothers with
Hearing deficit 585 1 unknown age or unknown LMP. Since other than
IQ <70 825 1
Cerebral palsy 97 II
white, black or Puerto Rican mothers were entirely
Epilepsy 170 II underrepresented in the study, the potential for
matched controls was either limited or nonexistent
in many hospitals, and they also were excluded.
Mothers who gave birth to singleton children with Gestation at registration was determined as the
each of the abnormalities were matched with women difference between the first day of the LMP and the
whose singleton children had none of the specified ab- calendar date of registration. The median gestational
normalities. The control mothers were selected by mat- age at registration was 17-20 weeks for the white
ching each case mother with a control from the same women and 21-24 weeks for black women in the
hospital, of the same race, with age ± 5 years, with NCPP, only 6% registering before the tenth week of
last menstrual period (LMP) ± 4 weeks, and early and gestation. The distributions of the gestational age of
late serum specimens taken at comparable times during the child at the time the early bloods and the late
pregnancy ( ± 6 weeks). bloods for this study were drawn, are given in Table 2.
Single controls were picked for each of the case Almost all of the late bloods were drawn between 37
mothers from those within the limits above using an weeks and 4-12 weeks post-delivery. Mothers whose
algorithm which ranked the 'closeness' of each poten- pregnancies ended with abortions, stillbirths or very
tial control to its case mother. For each case mother low birthweight babies had their late bloods drawn at
 PERINATAL TORCH INFECTIONS AND CHILDHOOD ABNORMALITIES 287

TABLE 2 Distribution of time of early and late blood samples in relation to last menstrual period (LMP) and termination of pregnancy

Early blood sample Late blood samplea

Time since Time of

LMP (weeks) No. % late blood No. °7o

0-12 942 16.2 At delivery (37-43 weeks) 1652 28.4

13-20 2559 43.9 Post delivery (4-12 weeks) 3865 66.4
21-28 2261 38.8 Other 306 5.3
>29 65 I.I
Total 5827 100.0 Toial 5827 100.1

a
Women whose pregnancy terminated in an abortion, stillbirth or a very low birthweight baby are not included.

earlier times, and are not included in this Table. The Briefly, they consisted of infected tissue cultures show-
cohort is such that we are able to assess the impact ing 3 + to 4 + CPE and processed as cell pack antigens
of infections essentially on the second and third using centrifugation and freeze/thawing procedures.

Downloaded from ije.oxfordjournals.org by guest on January 6, 2011

trimesters of pregnancy.
Materials and Methods of Laboratory Testing
The 'TORCH' agents were selected for this initial
study because of their established role as perinatal
infections which cause morbidity and mortality in
children. The Indirect Haemagglutination (IHA) and
Complement Fixation (CF) with Microtechnique were
used for this study. Previous studies utilizing these
systems for serological diagnosis 'en masse' demon-
strated the useful applications of these techniques. 67
Methods of Preparation of Antigens
The antigens used, their host-cell source and the strain
are shown in Table 3. Antigens were stored at -70°C.
A control antigen consisting of uninfected tissue and
fluids maintained in the same way as the antigen was
provided for each infectious agent.
Toxoplasma antigen was prepared from Toxo-
plasma gondii organisms obtained from peritoneal
exudates of infected mice. The organisms were rup-
tured by freeze-thawing once, followed by sonication
for 20 seconds at 4°C. The cell debris was removed by
centrifugation at 10000 g for 30 minutes. The superna-
tant fluid was stored as antigen.
Complement fixing antigens
Rubella antigen for Phase I of this study was prepared
with BHK-21 infected tissue cultures in a 5% suspen-
sion according to the methods reported previously. l0
Antigen for Phase 2 was prepared in Vero Tissue using
a 10% suspension of tissue culture material and fur-
ther purification with glycine hydroxide, freezing and
thawing three times, sonication for 3 minutes and
alkaline-extraction at 37°C for 6 hours.

Other Reagents
TABLE 3 Antigens used in NCPP study All control positive and negative sera were obtained
from personnel and remained the same throughout the
Antigen Host-cell source Strain
study.
The commercially prepared complement was anti-
Indirect haemagglulinaiion test
Cytomegalovirus MA-184 AD-169 body free of a screen of viruses including rubella. Each
Herpes 1 MA-196 Maclntyre lot was pretested before purchase to maintain a con-
Herpes II MA-196 MS stant titre.
Toxoplasma Mouse peritoneal fluid Beverly Sheep red blood cells were from one of two sheep
Complement fixation
Rubella (Phase 1) BHK 21 Gilchrist
chosen for their similarity in reaction and maintained
Rubella (Phase II) Vero Therien at the NIH farm.

Consistency of Procedures and Reagents

Indirect haemagglutination antigens Throughout the testing programmes many techniques
Cytomegalovirus and Herpes I and II antigens were pre- were applied to assure consistent results. All person-
pared according to the methods reported previously.8-9 nel, biologists, microbiologists, biology laboratory
288 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

technicians and medical technicians, were trained in
Microtechnique by the same person. The Micro-
technique materials, i.e. plates, loops, etc. remained
constant throughout these tests.
The paired serum specimens were coded and
scrambled for testing. These specimens which were
shipped and stored at -20°C were thawed rapidly.
Initial dilutions of 1:4 for CF, 1:8 for IHA viruses and
1:32 for toxoplasmosis were held at 4°C during testing
and -20°C for storage, even if only overnight. This
minimized the anticomplementary effect in CF and
some nonspecific reactions in IHA. Repeat tests used
this initial dilution unless anticomplementary or
nonspecific reactions were noted. In that case a new
dilution was made from the stock vial.
Antigens, except toxoplasma were prepared under
contract by Microbiological Associates, (MBA) Inc.
Strains, host sources and preparation methods re-
Anticomplementary effect or non-specific reactions
noted with control antigen were interpreted as shown
in Table 4.
All CF results were read by the same person. The
IHA tests had three different readers.
Definitions
Two definitions of antibody responses were applied
to the serological data. The first or seroconversion oc-
curred if the early titre was negligible and the later titre
satisfied the values indicated in Table 5. Significant in-
crease in antibody occurred if the difference between
titres for early and late bloods indicated a specific rise
from a low titre. The observed frequency of these in-
fections among the control cases according to type of
antibody response is given in Table 5. The high rate of
rubella reflects the occurrence of the rubella epidemic
in 1963.

Downloaded from ije.oxfordjournals.org by guest on January 6, 2011

mained constant. Specificity and antigenicity were
pretested by MBA and our laboratories for agreement
before acceptance. A common source of complement,
haemolysin, red blood cells and positive and negative
antisera was maintained from a minimum of different
lots.
Reproducibility within twofold dilution was re-
quired for each specimen tested by CF and fourfold
by IHA. All sera were titrated in duplicate on the day
of testing. Any questionable data and all fourfold
(CF) and eightfold (IHA) or greater differences were
retested for explanation and verification.
The paediatric outcomes were defined as follows:
Fetal death: abortion or stillbirth.
Neonatal seizures: seizures occurring within the first
28 days of life.
Congenital heart disease: gross structural abnorma-
lity of the heart or intrathoracic great vessels that was
actually or potentially of functional significance."
Congenital malformations: major central nervous
system malformations and related skeletal conditions,
eye, ear, respiratory tract, thoracic, alimentary tract
malformations, abnormalities of liver, bile duct and
spleen, genitourinary conditions, tumours and skin

TABLE 4 Interpretation of anticomplementary effect or nonspecific reactions noted with control antigen

Anticomplementery or nonspecific Test Reactions'1

reaction Serum Dilution
Scrum dilution

1:4 1:8 1:4 1:8 1:16 1:32 1:64 1:128

1-4 3-4 X X X X X X
1-4 2 X X X X Accept Accept
1-4 1 X X Accept Accept *
1-4 tr X 4 b Accept * * *
3bX
1-4 si tr X Accept
4 0 X
3 0 X • • • •
2 0 X - » » •
1 0 X • • • • •
Ir 0 4 b Accept • • •
3bX
si ir 0 Accept

"Test results rejected (X) wiih various levels of anticomplementary effect or nonspecific reaction.
b
Reading of 4
 PERINATAL TORCH INFECTIONS AND CHILDHOOD ABNORMALITIES
TABLE 5 Criteria for seroconversion and significant increase in antibody
Observed
frequency
Antigen Seroconversion among controls
Early Late Rate/10 000
To.xoplasmosis 0 or 32a 128 18.9
Rubella 0 3=4 88.4 b
4a 3=8
Cytomegalovirus 0or8a 5*32 35.7
Herpes 1 " 10.8
Herpes 11 6.3
indicates trace active at indicated dilution.
Study period covers lime in which 1963 rubella epidemic occurred
malformations. All syndromes, including mongolism
and Herler's Syndrome, and Pompe's diseases, were
excluded from this classification if they were the only
condition present in a child. Congenital heart condi-
tions were treated separately and were not included in
the congenital malformation category. Minor malfor-
mations, other than polydactyly, syndactyly and
enlargement of thigh and leg, were also excluded.
Microcephaly and macrocephaly: A child was con-
sidered micro or macrocephalic if his fronto-occipital
head circumference normed for gestational age, sex
and race deviated more than 2 standard deviations for
the mean on two or more examinations between new-
born and 7 years (newborn, 4, 8, 12, 48 months and 7
years).
Hypotonia at 1 year: hypotonic with deep tendon
reflexes.
Delayed motor development: delayed milestones of
motor development at 1 year.
Fine or gross motor development: clinical impres-
sion of abnormal fine or gross motor development at 4
years.
Enlarged liver at 1 Year: liver size greater than 2cm.
below the costal margin in the right midelavular line or
consistency unusual.
Hearing deficit at 3 years: failed to respond to one
or more of the frequencies 500, 1000 or 2000 at 20 db.
in either ear.
IQ <70 at 4 years: The 1960 revision of the
Stanford-Binet Intelligence Scale, Form L-M was ad-
ministered as close to the fourth birthday as possible.
Cerebral palsy at 7 years: disabling chronic deficit
characterized by aberrant control of movement or
posture, appearing early in life and not the result of
recognized progressive disease. Children with major
malformations of the CNS and those with motor
289
Observed
Significant increase frequency
in antibody among controls
Early Late Rate/10 000
3=32 3:256 (8-fold) 7.3
3:4 3:16 (4-fold) 55.4
5=8 3=16 (4-fold) 185.0
• 20.0
21.9
deficits acquired through infection or injury after the
Downloaded from ije.oxfordjournals.org by guest on January 6, 2011
first month of life were not included.
Epilepsy: defined as two or more asymptomatic
nonfebrile seizures or one asymptomatic nonfebrile
seizure after the fourth year of life.
Statistical Considerations
The size of this study was constrained by the number
of cases with abnormal outcomes, by the number of
available sera within the specified time interval and by
available resources to assay the bloods for multiple
controls. For most of the antigens, the estimated rate
of infection in the general population was low
(0.06-0.9<7o, for seroconversion see Table 5) and the
smallest increase in risk that we could expect to detect
with reasonable power (say 80%) was, for most of the
outcomes, that of a relative risk of greater than five or
more. This varied for each outcome, depending on the
number of matched pairs available for analysis.
Modest increases in risk would not have been detected
for those outcomes with the smallest number of
matched pairs, and the essentially negative findings of
this study must be viewed with this in mind. The out-
comes, such as congenital malformations, with large
numbers of matched pairs, did have sufficient sample
size, even for the rarest infections to provide definitive
negative results. Outcomes for which totally inade-
quate numbers of cases were available (e.g. diabetes
mellitus, 15 cases; visual impairment at 1 year, 36
cases; cranial or peripheral nerve abnormality 78 cases)
were not presented in this report, although serological
testing was done in the event there was a dramatic in-
crease in risk.
The odds ratio (OR) was used as an estimate of
relative risk. l2 An OR close to unity is an indication of
no increased risk of the outcome associated with the
290 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
infection. Odds ratios greater than one indicate a
positive association of the infection with the risk of the
outcome. The OR in a matched case-control study is
computed as the ratio of the number of matched pairs
where the affected child was born to a mother with a
specified infection and the control child was not, to the
number of matched pairs where the control child was
born to a mother with a specified infection and the
affected child was not. The importance of the OR is
a function not only of its magnitude, but also of the
severity and prevalence of the outcome. A doubling of
risk (OR = 2) for fetal death would be considered more
important than the same OR for the outcome of
asthma. Similarly, a doubling of risk when the pre-
valence of the outcome is of the order of magnitude
of one in a million is not equivalent to a doubling of
risk when the prevalence is around one in a thousand.
In addition, the observed OR must be assessed along
with measures of its statistical significance and the
width of the confidence interval with which it is
estimated, both of which are affected by the sample
size of the study as well as the inherent variability of
the observations. The confidence interval for an
observed OR gives the precision with which one makes
an estimate. A wide interval indicates a large varia-
bility in the estimate.
There were over 170 case-control studies involving
one of the five antigens and one of the 17 outcomes. In
testing whether differences in rates of seroconversion
or significant increase in antibody associated with the
abnormal cases were statistically significant (as in-
dicated by a high OR), MacNemar's test for matched
pairs was utilized.12 Since there was such a large
number of tests involved, with some, but not all, of the
test comparisons specified prior to the onset of the
study, some of the tests could be significant by chance
alone. We did not adjust significance levels using
multiple comparisons techniques, which would lessen
the chance of observing some clinically important risk
factors.13 Rather, we chose to consider the significance
values as indicators for risk factors that are candidates
for confirmation in future studies. Nevertheless, to
provide some protection against this problem only
statistical tests with a tabled P value of P<0.01 are
considered to be positive results.
RESULTS
The OR was computed for each of the case-control
studies, indicating the magnitude of risk of each out-
come associated with the presence of each of the five
antigens. The largest and smallest OR are given in
Tables 6 and 7 for seroconversions and significant in-
creases in antibody respectively. None of these OR
were statistically significant at P<0.01. In addition to
the examination of the individual outcomes, a sum-
mary category for 'any abnormality' was presented
which includes all of the listed outcomes as well as
several other rare outcomes, not given in the Tables.
It is important to interpret the nonsignificant fin-
dings with regard to both the observed ORs and the
power of the testing procedure to detect a significant
result. For example, for seroconversions (Table 6), the
OR for 'any abnormality' are near unity with the only
exception of toxoplasma. The OR of 1.8 (95% con-
fidence interval (CI): 0.9-3.8) for toxoplasmosis was
not significantly different from one, and was associ-
ated with a test having a power of only 50% to detect a
relative risk of 2 or more.
In general, the power of the completed studies was
not very high for any of the uncommon individual out-
comes and was fairly reasonable for the more common
Downloaded from ije.oxfordjournals.org by guest on January 6, 2011
outcomes (e.g. congenital malformations, IQ <70).
Examples of clinically important but not statistically
significant OR where a more substantial sample size
may have provided significant results, were the
presence of seroconversion to toxoplasmosis and a
significant increase in antibody to rubella with OR for
fetal death of 6 (95% CI: 0.7-132) and 5 (95% CI:
0.6-113) respectively. The extremely wide confidence
intervals reflect in part the small sample size.
DISCUSSION
The patients in the NCPP Study of over 54000 preg-
nant women have provided a number of opportunities
to investigate the role of infections in the perinatal
period.14 In the present study, the possible importance
of several different infections in pregnancy was in-
vestigated with antibody determinations performed on
the serial serum specimens obtained from patients
whose children had a wide range of different abnor-
malities, and matched controls. The patients in this
study had a mean gestational age at registration of
17-20 weeks for white women and 21-24 weeks for
black women. Only 6% registered before the tenth
week of gestation. Thus, serological data were avail-
able primarily for the second and third trimester of
pregnancy. The antigens selected included agents of
the TORCH group which have been suspected to cause
fetal damage. The present study showed no clear
association between any of the antigens studied and
any specific damage to the child. These 'negative' find-
ings were consistent with the absence of frequent signifi-
cant effects due to these agents in the second and third
trimesters of pregnancies. The TORCH agents are
generally important in the first trimester or as in the
case of herpes, at the time of delivery. Given the low
 PERINATAL TORCH INFECTIONS AND CHILDHOOD ABNORMALITIES 291

TABLE 6 Largest" and smallest odds ratios (OR) for abnormal outcome related to seroconversion by antigen

Abnormal Toxoplasma Rubella Cytome- Herpes I Herpes II

outcome galovirus

Fetal death (abortions 6.0

or stillbirths)
Birthweight ^2000 g 3.0
Neonatal seizures
Neonatal death 3.0
Congenital heart disease
Congenital malformations
Bilirubin >17mg%
Microcephaly 2SD
Macrocephaly 2SD
Hypotonia 1 year 2.0
Delayed motor development 2.0 2.3
Fine/gross motor development 3.0
Enlarged liver 1 year 3.0

Downloaded from ije.oxfordjournals.org by guest on January 6, 2011

Hearing deficit 3 years
IQ <70 4 years 3.0 2.0
Cerebral palsy 7 years
Epilepsy 3.0
Any abnormality 1.4 0.9 0.8

^ n l y OR of 2 or more and 0.5 or less are shown here, with the exception of the category Any abnormality.
-^Indicates X matched pairs where there was seroconversion in the case and no seroconversion in the control. The OR is indeterminate but is con-
sidered to favour the hypothesis of an OR greater than 1.
- Indicates an OR of 0.5 or less.

TABLE 7 Largest * and smallest odds ratios (OR) for abnormal outcome related to significant increase in antibody by antigen

Abnormal Toxoplasma Rubella Cytome- Herpes I Herpes II

outcome galovirus

Fetal death 5.0

Birlhweight $ 2 0 0 0 g 3.0 3.0
Neonatal seizures
Neonatal death
Congenital heart disease
Congenital malformations
Bilirubin >17mg% 4.0
Microcephaly 2SD 3.0
Macrocephaly 2SD 5.0
Hypotonia I year
Delayed motor development 1.9
Fine/gross motor development
Enlarged liver I year
Hearing deficit 3 years
IQ <70 4 years 4.0
Cerebral palsy 7 years 3a
Epilepsy 3.0
Any abnormality 1.2 I.I 1.0 1.1 1.1

a
Indicates X matched pairs where there was significant increase in antibody (as defined in Table 5) in the case and no significant increase in
antibody in the control. The odds ratio is indeterminate but is considered to favour the hypothesis of an odds ratio greater than 1.
- Indicates an odds ratio of 0.5 or less.
b
Only OR of 2 or more and 0.5 or less are shown here, with the exception of the category Any abnormality.
292 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
prevalance of herpes in this population, the absence of
an association with fetal damage may be due to the
small sample size.
A similar low prevalence of infection was also seen
for toxoplasmosis. Our recent report on toxoplasmosis
in 23000 pregnancies showed associations between
maternal infections and certain maternal and
paediatric findings.3 Several of the associations with
paediatric outcomes found were also examined but not
confirmed in the current study, and included hearing
impairment, microcephaly and 1Q. Several reasons
may account for the lack of confirmation. The mor-
bidity was defined differently for each of the outcomes
in the two studies. For example, IQ was measured at 7
years for the former study and 4 years here, and hear-
ing impairment was defined as bilateral deafness in the
former study and hearing deficit in either ear in this
study. The former study also examined primarily high
antibody to toxoplasma based on a single blood
sample at delivery or within 6-15 weeks postpartum,
not a predetermined high increase or seroconversion as
in the present study. The most likely reason for the
apparent inconsistency of the two studies is the more
sensitive study design of the first study. The former
cohort study involved the examination of some 23 000
women and the association of the outcomes with any
increases in titre to toxoplasmosis.
ACKNOWLEDGMENTS
The Collaborative Study of Cerebral Palsy, Mental
Retardation, and other Neurological and Sensory
Disorders of Infancy and Childhood is supported by
the National Institute of Neurological Disorders and
Stroke. The following institutions participated: Boston
Lying-in Hospital; Brown University; Charity
Hospital, New Orleans; Children's Hospital of Buf-
falo; Children's of Philadelphia; Children's Medical
Center, Boston; Columbia University; The Johns
Hopkins University; Medical College of Virginia; New
York Medical College; Pennsylvania Hospital; Univer-
sity of Minnesota; University of Oregon; University
of Tennessee; Infectious Diseases Branch, Biometry
and Field Studies Branch, and the Developmental
Neurology Branch, National Institute of Neurological
Disorders and Stroke.
We thank Alan Talbert, who did the computer pro-
gramming for this study.
REFERENCES
' Sever J L, Larsen J W Jr. and Grossman J H. HI. Handbook of
Perinatal Infections. Boslon: Little, Brown and Company,
1979.
2
Sever J L, Nelson K B and Gilkeson J R. Rubella epidemic 1964:
Effect on 6000 pregnancies. Am J Dis Child 1965; 100:
395-406.
3
Sever J L, Ellenberg J H, Ley A C, el al. Toxoplasmosis: Maternal
and Pediatric Findings in 23000 Pregnancies. Pediatrics 1988;
82: 181-92.
4
Niswander K, Gordon M. The Women and Their Pregnancies.
National Institute of Health Publication, 73-379. US Depart-
ment of Health, Education and Welfare, 1972.
Downloaded from ije.oxfordjournals.org by guest on January 6, 2011
5
Myrianthopoulos N C, French K S. An Application of the US
Bureau of the Census Socioeconomic Index to a Large, Diver-
sified Patieni Population. Soc Sci Med 1968; 2: 283-99.
6
Sever J L. Application of a Microtechnique to Viral Serological
Investigations. J Immunol 1962; 88: 320.
Sever J L, Huebner R J, Castellano G A, el al. Serologic Diagnosis
"en masse" with Multiple Antigens. Am Rev Resp Dis 1963;
88: Part 2.
Fuccillo D A, Moder F L, Traub R, et al. Micro Indirect Hemag-
glutination Test for Cytomegalovirus. Appl Microbiol 1971;
21: 104.
Fuccillo D A, Moder F L, Catalano L W Jr. el al. Herpes Hominis
Types 1 and II: A Specific Micro Indirect Hemagglutination
Test. Proc Soc Exp Biol Med. 1970; 133: 735.
10
Sever J L, Huebner R J, Castellano G A, el al. Rubella Comple-
ment Fixation Test. Science 1965; 148: 385.
" Mitchell S C, Korones S B and Berendes H W. Congenital Heart
Disease in 56109 Births - Incidence and Natural History. Cir-
culation 1971; 43: 323.
12
Fleiss J L. Statistical Methods for Rates and Proportions. New
York, John Wiley and Sons, 1973.
13
Mantel N, Haenszel W. Statistical aspects of the analysis of data
from retrospective studies of disease. JNCI 1959; 22: 719-48.
14
Sever J L. Infections in pregnancy: Highlights from the col-
laborative perinatal project. Teratology 1982; 25: 227-37.
(Revised version received June 1991)