You are on page 1of 8

Antecedents of Cerebral Palsy

I. Univariate Analysis of Risks


Karin B. Nelson, MD, Jonas H. Ellenberg, PhD

\s=b\ A large prospective study inves- relate to risks for cerebral palsy, as was not the result of recognized progres¬
tigated prenatal and perinatal ante- evaluated in univariate analyses. A sive disease. Children with meningomyelo-
cedents of chronic motor dysfunction companion article will consider ante¬ celes and those with motor deficits ac¬
(cerebral palsy [CP]), evaluating approx- cedents of cerebral palsy (CP) and of quired through infection or injury after the
imately 400 characteristics of the moth- its clinical subtypes, employing multi¬ first month of life were not included. Unlike
ers, pregnancies, or deliveries. In addi- variate techniques.
several previous reports from the NCPP,
tion to confirming some, but not all, of this report includes mild, as well as moder¬
the classic risk factors for CP, this study ate and severe, CP.
observed relatively large increases in SUBJECTS AND METHODS The statistical test used to evaluate four¬
the CP rate in association with maternal fold contingency table assessments of di-
mental retardation, seizure disorders, Approximately 54,000 pregnancies were chotomous antecedents for CP for signifi¬
hyperthyroidism, or with the administra- studied at 12 cooperating university hospi¬ cance was the 2 with 1 df (using a
tion of thyroid hormone and estrogen in tals between 1959 and 1966. The study continuity correction), except where other¬
pregnancy. Some risk factors were pre- sample was approximately 46% white and wise indicated. Fisher's exact test was used
dictive of CP only insofar as they were 46% black; women of Hispanic descent where cell size warranted it. Polychot-
associated with low birth weight or low made up a majority of the remaining sub¬ omous antecedents were analyzed for glo¬
Apgar scores. Among factors not signifi- jects. The socioeconomic level of the NCPP bal significance using the R x C contin¬
cantly related to CP rate were maternal population was somewhat lower than for gency table format. Fourfold comparisons
age, parity, socioeconomic status, the total population of the United States as of subgroups within larger contingency ta¬
smoking history, maternal diabetes, first assessed by the Bureau of the Census.1 bles were made only when the 2 test for the
trimester vaginal bleeding, kidney or Information relating to the sample and to larger table was significant at the .05 level.
bladder infection, moderate hyperten- study procedures is available elsewhere.z'3 Given the many 2 tests made within this
sion, long cord, use of anesthetic At the initial prenatal visit, medical, large dataset, some were likely to achieve
agents, or use of oxytoxics for initiation obstetric, and family histories were re¬ nominal levels of statistical significance by
or augmentation of labor. Duration of corded. At each subsequent visit, medical chance alone. We did not adjust signifi¬
labor, whether precipitate or prolonged, and obstetric events and medication intake cance levels using multiple comparison
was not a risk factor for CP. were noted. Patients were examined on techniques, which would lessen the chance
(AJDC 1985;139:1031-1038) admission in labor, and the course of labor of observing some clinically important risk
and delivery was observed.3 After delivery, factors.5 Rather, we chose to consider the
rPhe Collaborative Perinatal Project a physician reviewed all prenatal and labor significance values simply as indicators for
of the National Institute of Neu¬ and delivery records, including documenta¬ risk factors that are candidates for confir¬
tion of medical and surgical conditions of mation in future studies.
rological and Communicative Disor¬ the gravida before or during the pregnancy. Most continuous variables were exam¬
ders and Stroke (NCPP), a large pro¬
Ascertainment of drug exposure and its ined in both dichotomous and polychot-
spective study, was undertaken to timing was as described by Heinonen et al,4 omous forms. Intervals were chosen a pos¬
investigate the association of maternal except that we considered drug exposure teriori, based on observed risk. Evaluation
and pregnancy conditions and events for the entire pregnancy. Neonatal infor¬ of possible linear trends in the risk of CP for
of labor, delivery, and the neonatal mation was derived from delivery room and polychotomous variables used weighted
period, with chronic neurologic disor¬ nursery records. least-squares regression analysis of the ob¬
ders of childhood. This article reports Examinations in the first year of life were served probabilities of risk, with weights
the major results of the NCPP as they the basis for the recognition of congenital determined by the number of patients at
malformations. A pediatric and neurologic each observation point.
examination performed when children To examine whether variables signifi¬
were 7 years old provided the basis for cantly related to CP for the total population
From the Developmental Neurology Branch
(Dr Nelson), and the Biometry and Field Studies diagnoses of CP. Children with a CP diag¬ were operative solely or chiefly through
Branch, Intramural Research Program (Dr El- nosis at 1 year of age who died before 7 their association with low birth weight, we
lenberg),National Institute of Neurological and years of age were included. Cerebral palsy evaluated each variable significant for all
Communicative Disorders and Stroke, Bethesda,
was defined as a chronic disability charac¬ birth weights, for relationship to CP within
Md.
Reprint requests to 7550 Wisconsin Ave, Room terized by aberrant control of movement or the 91% of the population with birth
8C-04, Bethesda, MD 20205 (Dr Nelson). posture that appeared early in life and that weights over 2,500 g.

Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 06/21/2015


A history of maternal diabetes,
Table 1.—Risk Factors Ascertainable Prior to Pregnancy or at First Prenatal Visit*
length of time to become pregnant,
Relative Risk and maternal and sibling rhesus blood
Observed Rate Total
groupings were not associated with
Condition Prevalence, % Cerebral Palsy, % Birth Weight >2,500 g CP. Also not predictive of CP were
Demographic prior infertility, maternal sickling, age
Race: white at menarche, season of last menstrual
Mother working period, gonorrhea, syphilis, psychosis
Maternal medical or neurosis prior to pregnancy, mental
conditions before illness or congenital malformations in
pregnancy
Mental retardation 6.5t 10.3* family members, Roentgen ray expo¬
sure of the abdominopelvic area,
Seizures, recurrent,
major, active 0.4 2.7 6.7t 8.6* thrombotic illness or coagulation
Seizures, any active 0.8 1.4 3.5t 4.5* defect, glomerulonephritis, tuberculo¬
Hyperthyroidism 0.6 2.0 4.9f 4.7* sis, leiomyomata, neurologic disorders
Prior hospitalization 13.5 0.6 1.6* 1.2 other than seizures or retardation,
Reproductive history neuromuscular or gastrointestinal
Menstrual cycle tract disease, asthma, or organic heart
Long 3.6 0.8 1.9* 2.4* disease, including rheumatic fever
Unusual 6.9 0.7 1.7* 2.2* prior to pregnancy.
Last pregnancy None of the relative risks (RRs)
Birth weight <2,000 g 3.9 1.0 27* 0.5
Neonatal death 1.4 1.3 3.1* 2.0
relating to social and demographic
characteristics exceeded 2, and the
No. of fetal deaths -2 0.6 1.4 3.4* 3.1 observed rate of CP associated with
History of prior siblings any of these factors was below 1%. For
Motor deficits 1.2 1.5 3.8* 2.7
most medical factors ascertainable be¬
Sensory deficits 1.0 1.1 2.8* 1.6
fore or very early in pregnancy, RRs
Mental retardation 1.5 1.1 2.7* 2.9* were in the range of 2 to 10. Babies
*For each factor, the computed 2 for the comparison between the observed rate of cerebral palsy for born to mothers who were mentally
pregnancies with the factor as compared with those without is less than the tabled value of 2.1 The
computation of relative risk for conditions assuming prior pregnancies was computed as the rate of retarded or had seizures were at high¬
cerebral palsy in the group with the factor over the rate of cerebral palsy in the group without the condition est risk, with absolute rates of CP
or no prior pregnancy.
*P<.01. approaching 3%.
*P<.05.
The Pregnancy
RESULTS pregnancy spacing, smoking history Noneclamptic seizures during preg¬
Of 51,285 pregnancies that produced (number of cigarettes per day and nancy, not symptomatic of acute en¬
live-born singleton infants, outcome number of years smoked), and inter¬ cephalopathy, were associated with a
course frequency were not associated relatively high RR (Table 2).
was known at 7 years of age for 45,559
with significantly increased risk of CP, Some of the clinical elements of tox¬
children, of whom 189 had CP. nor were maternal height or prepreg¬ emia were associated with increased
The Mother nant weight. Early registration for risk if they were present in a severe
A higher rate of CP was observed in prenatal care was not associated with a degree. Severe proteinuria in the sec¬
white children, and a lower rate was lower rate of CP, nor was late registra¬ ond half of pregnancy, with 5 g or more
noted in children whose mothers were tion with increased risk. of protein present in the urine per 24
employed outside the home (Table 1). Maternal mental retardation (as rec¬ hours, was a substantial risk factor for
There was a slight and nonsignificant ognized clinically and obviously under- CP; moderate proteinuria (early or
increase in risk in the lowest socioeco¬ ascertained), seizure disorders6 (espe¬ late) and severe proteinuria in the first
nomic status group, as defined by oc¬ cially if recurrent and generalized), half of the pregnancy were not associ¬
cupation and education of the head of and hyperthyroidism prior to preg¬ ated with CP. High blood pressure
the household and total family in¬ nancy were significantly associated (diastolic pressure, -110 mm Hg) in
come.1 This group included 3% of white with CP in the offspring. There was an the third trimester was associated
grávidas and 12% of black mothers. increase in the risk of CP with an with a doubled risk of CP. High blood
Grávidas 14 years of age and less, a increased number of confining ill¬ pressure earlier in pregnancy was un¬
very low prevalence group and one nesses in the year before the last men¬ common, and was associated with a
whose members were also at the low strual period (trend significant, doubling of risk that was not statis¬
end of the socioeconomic spectrum, _"<.05). Menstrual cycles longer than tically significant. Moderate hyper¬
contributed a slight and nonsignificant 36 days, as well as unusual or irregular tension (diastolic pressure 90 to
excess of risk. Maternal education, cycles, were associated with some in¬ 110 mm Hg), whether early or late in
marital status, parity, paternal age, crease in the risk of CP. pregnancy, low blood pressure in the

Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 06/21/2015


fewer visits. It appeared that it was
Table 2.—Medical Conditions and Hormone Treatment During Pregnancy the low birth weight associated with
Relative Risk* early termination of pregnancy, and
therefore lesser opportunity to keep
Observed Rate Total
Condition Prevalence, % ot Cerebral Palsy, % Birth Weight >2,500 g prenatal appointments, that might ac¬
Conditions in pregnancy count for the relationship of fewer pre¬
Seizures, noneclamptic 0.4 1.8 4.3+ 6.8+ natal visits with CP.
Tonic-clonic, Rubella and toxoplasmosis were
0.1 2.9 7.2* 11.7+
recurrent
rarely diagnosed by clinical, nonsero-
Heart disease, logic criteria in pregnancy, and the
asymptomatic 0.9 1.3 3.1* 3.0
increase in risk associated with clinical
Severe proteinuria, late 1.1 1.6 3.9* 5.1 +
Severe hypertension,
diagnosis in the mothers was not sta¬
third trimester 2.9 0.8 2.1* 2.2 tistically significant.
Incompetent cervix 0.3 2.4 5.9* 0.0
Maternal diabetes in pregnancy was
not found to be a risk factor for CP, nor
Bleeding, third
trimester 13.6 0.7 1.8+ 1.6* were hyperthermia, kidney or bladder

Polyhydramnios 1.3 1.0 2.5* 3.2* infection, asthma, leiomyoma and


Hormones other gynecologic tumors, psychosis
Estrogen 0.8 1.4 3.3* 1.1 or neurosis, air travel, sick pet in the
Progesterone 1.5 1.1 2.7* 2.0 home, Roentgen ray exposure, jaun¬
Thyroid 1.4 0.9 2.3* 2.3 dice, vomiting including hyperemesis
Thyroid and estrogen 0.1 3.3 8.0* 6.2 gravidarum, rheumatic fever, clin¬
No. of prenatal ically diagnosed syphilis, herpes
visits -6 30.8 0.6 1.7+ 1.0 simplex, mumps or measles during
*For each factor, the computed 2 for the comparison between the observed rate of cerebral palsy for pregnancy, hyperthyroidism and
pregnancies with the factor as compared with those without Is less than the tabled value of 2.' hypothyroidism in pregnancy, and
+P-C.01. sulfa
*P<.05. treatment with antibiotics,
drugs, diuretics, anorectic agents,
vaginal contraceptives, antihyperten-
sive agents, and barbiturate and non-
middle trimester, and edema of the term birth weight; apparently the cor¬ barbiturate hypnotic agents and anti-
face or hands were not related to CP relation of low weight gain with low con vulsant medications.
rate. birth weight in the child accounted for The RRs for pregnancy characteris¬
Incompetent cervix in the study the association. tics were in the range of 2 to 12. The
pregnancy, although rare, had one of A heightened risk of CP was ob¬ associated absolute CP rates for most
the highest RRs of obstetric conditions served in children of women who re¬ factors ranged between 1% and 2%,
examined, with that risk limited to ceived estrogenic hormone, progester¬ and never exceeded 3%. Except for
infants with low birth weights. Vaginal one, or thyroid hormone, or thyroid third trimester bleeding, pregnancy
bleeding in the third trimester was hormone and estrogen, a rare combi¬ factors significantly predictive of CP
relatively common, and was signifi¬ nation, in pregnancy. This observation were uncommon characteristics.
cantly associated with CP. Vaginal was investigated in further detail (see
The Labor and Delivery
bleeding in the first trimester was not below).
associated with any increase in the The average number of prenatal vis¬ Vaginal bleeding present at the time
observed rate of CP, and second tri¬ its per gravida was nine. Offspring of of admission to the hospital for deliv¬
mester bleeding was associated with women who made six or fewer prenatal ery was associated with an increased
an RR of 1.5, a nonsignificant increase. visits had a slightly higher rate of CP risk of CP, as were diagnoses of ab-
Maternal anemia, even if extreme than children of women who made ruptio placentae, placenta previa, and
(lowest hemoglobin level, <7 g/dL), seven or more visits; this was true for marginal sinus rupture (Table 3). In
was not significantly associated with a the total population, but not for babies weighing over 2,500 g, bleeding
risk of CP. Polyhydramnios was associ¬ women whose babies weighed over on admission, abruptio placentae, and
ated with a trebling of the risk of CP; 2,500 g. Of the infants with low birth marginal sinus rupture were not sig¬
neither central nervous system mal¬ weights, 53% were born following nificantly related to CP.
formations nor anomalies preventing pregnancies in which there had been Rupture of membranes more than
swallowing were observed in affected six or fewer visits, and 75% of the CP in 24 hours prior to delivery was asso¬
infants. The risk of CP was inversely children with low birth weights oc¬ ciated with a tripling of the risk of CP
related (significant linear trend, curred in this subgroup. In contrast, in the total population, but was not
P<.05) to maximum maternal weight no excess of CP was seen in large related to CP among babies weighing
gain in pregnancy, but this relation¬ infants, 28% of whom were born after over 2,500 g. Chorionitis, identified by

ship was not present in children of pregnancies in which there were six or marked neutrophilic infiltration of the

Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 06/21/2015


Table 3.—Risk Factors in Labor and Delivery reported.8,9 For children with low birth
weights, the observed rate of CP was
Relative Risk* higher in white than in nonwhite in¬
fants (Figure).9 Gestational age, pla¬
Observed Rate Total
Condition Prevalence, % of Cerebral Palsy, % Birth Weight -2,500 g cental weight, and body length were
Vaginal bleeding at related to CP in the total population,
admission 5.6 0.8 2.2* 1.0 but all were correlated with birth
Placental complications 3.2 1.5 3.9* 2.4* weight and were not associated with
Previa 0.6 2.0 4.9* 6.2* CP in babies weighing over 2,500 g. A
Abruption 1.6 1.4 3.5* 2.0 nonsignificant association was noted
Marginal sinus between the presence of a single um¬
rupture 1.2 1.1 2.7* 0.8 bilical artery and CP rate. Short cord
Rupture of membranes (<40 cm) was a risk factor for CP.
>24 hr before
delivery 5.4 1.0 2.8* 1.0 Neither size of placental infarcts over
3 cm, nor infarct number, nor long
Chorionitis 2.3 1.8 4.8* 2.6*
umbilical cord (-80 cm) were related
Presentation
Breech 2.7 1.8 4.6+ 3.9* to CP risk.
Face, brow, Delayed onset of respiration and
transverse 0.5 2.1 5.1 + 5.3* first cry were associated with an in¬
Occiput anterior 10.2 0.4 0.4+ 0.6 creased risk of CP; for neither event
Delivery mode was a one- or two-minute delay accom¬
Breech 2.3 1.9 4.9+ 3.4+ panied by a marked increase in risk.
Vertex 92.8 0.4 0.4+ 0.5+ A series of ratios relating physical
Lowest fetal heart measurements at birth was used to
rate -60 beats/min 1.3 1.8 4.5+ 2.9
2.0 2.6
explore intrauterine growth patterns
Hemorrhagic shock 0.3 5.0+ as predictors of CP. A low weight for
*For each factor, the computed 2 for the comparison between the observed rate of cerebral palsy for
pregnancies with the factor as compared with those without is less than the tabled value of 2.1 body length at birth was associated
+P<.01. with somewhat heightened risk in the
*P<.05. total population, but not in the babies
weighing over 2,500 g. The ratio of
placental chorion on microscopie in¬ risk of CP at either the precipitate or head circumference to birth length
spection, and breech, face, brow, the prolonged ends of the distribution. demonstrated a slight increase in risk
transverse, and other presentations, Length of labor was not different in among infants whose heads were large
and breech deliveries, were related to births of children with spastic CP vs for their body length. For the total
CP risk. Delivery by emergency ce¬ controls in a previous study.7 birth weight group a high placental
sarean section was associated with an Bipartite placenta and continuous weight for birth weight was associated
RR of 1.3, which is not a statistically gas, caudal, or spinal anesthesia were with some increase in the risk of CP,
significant increase in risk. Cesarean not significantly related to CP. Also but no such relationship was observed
sections performed because of previ¬ not significantly predictive of CP were among large infants; smallness of the
ous operative deliveries and labor and cord prolapse, nuchal cord, cord placenta for birth weight was not re¬
delivery drugs were not significant around body, midforceps delivery, and lated to CP. None of the groups identi¬
predictors of CP. meconium staining of the amniotic fied through these ratios of intra-
A lowest fetal heart rate below fluid, although for all of these except uterine growth measures had a high
60 beats per minute, as recognized nuchal chord there were positive asso¬ absolute rate of CP.
by auscultation, was a predictor of ciations with CP. (In the small group of The risk of CP was increased four¬
CP, but a lowest rate between 61 and children with some of these complica¬ fold for a child with one or more major
80 beats per minute was not. Highest tions and low Apgar scores at five malformations outside the central ner¬
fetal heart rate was not significantly minutes, there were significant associ¬ vous system (as classified by Myri-
related to risk of CP. The obstetricians' ations with CP.3) anthopoulos and Chung,10 with slight
judgment of arrested progress of labor For characteristics of labor and de¬ modification). Minor malformations in
or uterine dysfunction was not a signif¬ livery, RRs ranged between 2 and 5, children free of major non-central
icant predictor of CP, and oxytocin and with absolute CP rates of 2% or less. nervous system malformations were
other medicinal agents used for initia¬ not associated with an increased rate
tion of labor or for augmentation, and DeliveryRoom Observations ofCP.
induction of labor by amniotomy, were Male infants developed CP more Characteristics of the infant once he
not. Duration of the first stage of labor often than female infants did. Birth or she could be examined directly were
to more than 40 hours, or the second weight and Apgar scores (Table 4, stronger predictors of CP than the
stage to more than 129 minutes, were Figure) were important predictors of prenatal and obstetric factors that we
not associated with differences in the CP in this population, as has been studied.

Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 06/21/2015


Table 4.—Characteristics of Children at Birth
was an important risk factor for CP.
This finding is consistent with other
Relative Risk work describing the neurologic mor¬
Observed Rate Total bidity stemming from symptomatic in¬
Condition Prevalence, % of Cerebral Palsy, % Birth Weight -2,500 g fants in that majority of the newborn
Noted in delivery room population with birth weights over
Male 50.5 0.5 1.6* 1.7* 2,500 g.U12
Birth weight Respiratory problems in the nur¬
-2,500 g 9.2 1.5 4.9*
sery period in general, and those re¬
-2,000 g 2.0 4.5 13.7* lated specifically to respiratory dis¬
Placental weight tress syndrome and to aspiration,
<325 g 10.1 1.1 3.2* 1.0
were predictors of CP. The severity of
Time to cry _5 min 1.2 3.8 11.2* 10.8*
Time to breathe >3 min 0.7 3.5 9.2* 12.5*
respiratory problems, whatever the
specific diagnosis, was related to risk
Apgar score 0-3+ of CP.
1 min 4.8 1.9 6.4* 5.2*
5 min 0.9 6.5 20.8* 20.0*
Anemia in the infant with a lowest
hematocrit below 40% was a predictor
Major congenital of CP; hematocrit values over 70%
malformation, non-central
nervous system 6.8 1.3 3.8* 3.9* were not. A peak bilirubin level over
*For each factor, the computed 2 for the comparison between the observed rate of cerebral palsy for 16 mg/dL was a risk factor for CP.
pregnancies with the factor as compared with those without Is less than the tabled value of 2 at the P< .01 Infections of various sites, definite
level.1
+The computation of relative risk for children with low Apgar scores was computed by comparison with and suspect, and septicemia and men¬
a score of 7 to 10 in each time group. ingitis, both very uncommon, were
associated with CP. Maternal postpar¬
Table 5.—Characteristics of Children in Newborn Nursery tum endometritis was not related to
CP risk.
Relative Risk*
Observed Rate
An observation of interest was the
Condition Prevalence, % of Cerebral Palsy, % Total Birth Weight 2,500 g fairly high RR associated with the
In incubator _3 days+ 1.1 5.4 22.4* systemic administration of antibiotics
Newborn seizures 0.3 19.8 53.6* 63.3* in the newborn nursery. Since many
Diminished cry 1.2 5.1 14.5* 12.9* infants to whom antibiotics were given
Diminished activity 1.9 4.8 were not recorded to have definite or
14.6* 13.6*
Respiratory problems suspected sepsis or sites of infection,
(any) 1.5 3.4 9.2* 7.7* this observation was pursued, and will
Respiratory distress be discussed below.
syndrome 0.2 8.8 21.9* 49.3* Overall, the highest relative risks
Aspiration 0.2 10.3 25.7* 27.0* for CP observed in the prenatal, in¬
Lowest hematocrit <40% 2.5 3.7 10.9* 6.2* trapartum, and neonatal course ac¬
High bilirubin >16 mg/dL 2.5 2.1 5.5* 2.8§ companied very low Apgar scores, sei¬
Infection/treatment zures, and respiratory distress in large
Infection (any
definite) 1.2 2.6 6.5* 3.4 infants, all of which are characteristics
Infection (any recognized only after birth.
suspect) 1.0 3.2 8.5* 5.7*
Meningitis (definite) 0.02 12.5 30.5§ 0.0 Thyroid, Estrogen,
and Associated Factors
Meningitis (suspect) 0.02 11.1 27.1§ 0.0
Septicemia (definite) 0.06 7.1 17.4* 0.0
The observed rate of CP was in¬
Septicemia (suspect) 0.3 2.7 6.6§ 10.1§ creased in children whose mothers re¬
Antibiotics
ceived thyroid hormone or estrogen in
10.5 1.5 5.0* 3.3*
*For each factor, the computed 2 for the comparison between the observed rate of cerebral palsy for
pregnancy (Table 2). Approximately
pregnancies with the factor as compared with those without Is less than the tabled value of 2.1 two thirds of the women who received
+Term weight infants only. one or both these agents had hyper-
*P<.01. thyroidism or hypothyroidism, diabe¬
§P<.05.
tes, polyhydramnios, or a prior preg¬
The Nursery Period zures were all important risk factors nancy loss. In some women thyroid
for CP. hormone was prescribed for weight
A number of newborn signs had rela¬ A need for special nursing for in¬ reduction or as "pep pills", and in some
tively high RRs (Table 5). Evidence of fants weighing over 2,500 g, reflected women itwas included as part of a
neonatal depression, altered homeo- here by incubator care lasting for regimen intended to prevent preg¬
static mechanisms, and neonatal sei- three days or longer in large infants, nancy loss. In women who bore chil-

Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 06/21/2015


amined the indications, agents, and
birth weight as predictors of CP in
stepwise logistic-regression analyses.
Variables describing thyroid disease,
prior fetal death, polyhydramnios, and
diabetes and its features (duration
longer than five years, insulin use, and
episodes of ketoacidosis, diabetic
coma, and insulin reactions) were in¬
cluded. Birth weight less than 2,500 g,
polyhydramnios, estrogen, and the
joint administration of estrogen and
thyroid hormone were identified as
significant by the logistic analyses;
diabetes and its complications were
not. None of the 206 offspring of dia¬
betic women who did not receive hor¬
mones or have polyhydramnios or thy¬
roid disease had CP. Although the
number of infants with hormone expo¬
sure and CP was small, these results
suggest that prenatal exposure to ex¬
ogenous thyroid hormone and estro¬
gen were associated with an increased
risk of CP independent of the indica¬
tions for their administration.
Thyroid hormone, estrogen, and
progesterone are all agents with im¬
portant effects on developing neuronal
systems.13"16 Estrogen administration
enhances the effects of thyroid hor¬
mone."
Antibiotics and
Related Factors
Neonatal sepsis or meningitis, re¬
<1,501 1,501-2,000 2,001-2,500 >2,500 corded as definitely present or only
Birth Weight, g suspect, were associated with high
RRs for CP (Table 5). However, these
W 54 192 1,038 19,350 conditions were very uncommon, and
NW 158 472 2,018 19,039 these and other neonatal bacterial in¬
fections were evident in the histories
of only a small proportion of children
with CP.
Birth weight groups and rate of cerebral palsy, in white and nonwhite children. At given low More than 10% of children in the
birth weight, observed rate of cerebral palsy was higher in white (solid line) than in nonwhite NCPP received systemic antibiotics in
(broken line) children. W indicates number of white children; NW, number of nonwhite the newborn nursery, while only 2% of
children.
the population had a definite or sus¬
pect recorded infection. Children who
dren with CP, thyroid and/or estrogen While some of the indications for the received antibiotics in the nursery but
therapy usually was begun in the third administration of thyroid hormone and who had no recorded infection were
month of pregnancy or earlier. estrogen were associated with an in¬ three times as likely to die in the first
The CP in children exposed in utero creased risk of CP, these agents ad¬ year, and survivors were four times as
to thyroid hormone, estrogen, or both ministered without these indications likely to have CP as were children who
was not homogeneous. One child had were associated with a trebling of risk. had neither infection nor antibiotics.
malformed kidneys, while in the other To assess further whether it was in¬ Neonatal seizures were seven times as
children there were no obvious malfor¬ dications for hormone administration common among children who received
mations. No child was recognized to or possible effects of the agents them¬ antibiotics but had no injection; the
have abnormal thyroid function. selves that were risk factors, we ex- order of occurrence of seizures and

Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 06/21/2015


antibiotic administration was not in¬ which CP was later the outcome as for most children with a given risk factor
vestigated. those subjects without CP. It was thus did not have CP. In addition, most
The antibiotics received by study possible to ascertain, with information conditions with high RRs were rare;
children were most frequently a com¬ recorded prospectively, the frequency the few prelabor conditions with RRs
bination of penicillin and strepto¬ of suspected antecedents of disease not exceeding 3 all had prevalences less
mycin. Infants treated with antibiotics only in cases of CP but, of equal impor¬ than 1%. The conditions with the high¬
who had no infection sometimes re¬ tance, in noncases. Review of histories est RRs, and with highest absolute
ceived these agents prophylactically of children with CP cannot provide rates of CP, were characteristics of
because of delivery in unsterile condi¬ reliable information on cause unless infants after their births.
tions, sometimes had hyperbilirubi- accompanied by consideration of com¬ Many factors found in the NCPP to
nemia and exchange transfusions, parable events, comparably ascer¬ have high RRs have long been consid¬
and, more often than infants who did tained, in unaffected children. ered antecedents of CP. Other condi¬
not receive antibiotics, were of low Perhaps the foremost of the limita¬ tions long assumed to be related to CP,
birth weight and/or had low Apgar tions of the NCPP, is its date. Ob¬ such as maternal diabetes, extremes of
scores. A factor associated with anti¬ stetric and neonatal care have changed maternal age, and low socioeconomic
biotic use without infection was the markedly since the NCPP was begun. status, were related only slightly, if at
presence of respiratory difficulty in Changes have been most dramatic in all.
the newborn period, whether ascribed the of the infant with low birth
care We have previously reported that
to respiratory distress syndrome, ate¬ weight. For this reason, we have some risk factors, especially complica¬
lectasis, or aspiration. Among infants stressed results relating to antece¬ tions of labor and delivery, were asso¬
free of infection, increasing severity of dents of CP in babies weighing over ciated with increased risk of CP only in
respiratory difficulty was associated 2,500 g, in whom much of the CP children who also had low Apgar
with increasing frequency of antibiotic occurs. scores.3 As the present report indi¬
use, from 8.9% in children with no As in any observational study, in the cates, other risk factors were predic¬
respiratory problems to 74% in those NCPP associations between antece¬ tive of CP only insofar as they were
with severe respiratory difficulties. dents and outcome could be estab¬ associated with low birth weight. In¬
It is probable that doctors pre¬ lished but the nature of the associa¬ competent cervix, low birth weight of
scribed antibiotics for an infant with¬ tions—whether causal or not—could an older sibling, and prior neonatal
out infection because of a perception of not be determined; to do that would death of a sibling, for example, were
the infant as ill. Antibiotic use may require a differently designed study not predictors of CP in babies who
thus serve as a flag for risk. In addi¬ capable of testing the effects of inter¬ weighed over 2,500 g at birth, al¬
tion, some antibiotics are neurotoxic, ventions, necessarily a second-genera¬ though they were predictive in the
penicillin being especially associated tion effort. On the other hand, this total population. Thus, both low Apgar
with seizures and streptomycin with study could, because of the large num¬ scores and low birth weight seem to act

ototoxicity.18 ber and variety of characteristics ex¬ as pathways for the expression of cer¬
Of children with CP, 37% received plored, expand the set of questions tain earlier pathogenetic factors.
antibiotics in the newborn nursery; in asked relating to the etiology of CP. A few factors or clusters of factors
24% of children with CP this was de¬ This study could thus generate hy¬ identified empirically rather than by
spite the absence of definite or sus¬ potheses on etiology but not provide prior hypothesis were found to have
pected infection. definitive tests of them. reasonably high RRs or to occur in a
This study could rule out associa¬ substantial proportion of cases of CP.
COMMENT tions insofar as sample size permitted. Of these, polyhydramnios and hor¬
In a recent monograph that re¬ A study that differed only by its larger mone administration in pregnancy,
viewed the literature on the etiology of size might have found statistical signif¬ and associated factors, raise tantaliz¬
CP, Alberman and Stanley19 discussed icance for some uncommon or weakly ing questions. Administration of thy¬
the paucity of methodologically ade¬ acting factors not found to be impor¬ roid hormone and estrogen in an effort
quate studies on the subject. As they tant in this study. However, such ante¬ to prevent early pregnancy loss is a
remarked, "The study of the cerebral cedents could not account for a large regimen no longer in use. Interest in
palsies presents epidemiologists with proportion of CP. whether these agents, given in preg¬
every possible pitfall." The NCPP was Some of the prenatal and perinatal nancy, might have contributed to
designed to evaluate the antecedents risk factors evaluated for CP were motor disorders in the offspring de¬
of CP, employing epidemiologie ap¬ indeed found to be fairly risky, in that rives from the more general possibility
proaches that aimed to avoid a few of an individual possessing such charac¬ that aspects of maternal endocrine sta¬
those pitfalls. teristics was at a risk of developing CP tus might in some cases be related to
The prospective structure of the severalfold greater than that for indi¬ childhood motor disorders. The find¬
NCPP allowed identification of viduals without the characteristics. ing that long or unusual menstrual
mother-child pairs before outcome was However, even conditions with fairly cycles were among the few maternal
known, and data were collected in the high relative risks were not associated characteristics predictive of CP is com¬
same manner for mother-child pairs in with high absolute rates of CP, and patible with this possibility.

Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 06/21/2015


The NCPP set out to test, quantify, tions may contribute to a broadening of sity, Baltimore; Medical College of Virginia,
and compare the risk of CP associated our present rather limited set of hy¬
Richmond; New York Medical College; Pennsyl¬
vania Hospital, Philadelphia; University of Min¬
with a large set of maternal, preg¬ potheses regarding the etiology of CP. nesota, Minneapolis; University of Oregon, Port¬
nancy, labor and delivery, and neonatal land; University of Tennessee, Memphis; and the
characteristics. In addition to confirm¬ Biometry and Field Studies Branch and the
The Collaborative Study of Cerebral Palsy, Developmental Neurology Branch, National In¬
ing some, but not all, of the classic risk Mental Retardation and Other Neurological and stitute of Neurological and Communicative Dis¬
orders and Stroke, Bethesda, Md.
factors for CP, this study has identified Sensory Disorders of Infancy and Childhood is
supported by the National Institute of Neu¬ We thank Charlotte Jackson, who did the com¬
several other characteristics of the puter programming for this study, Martha Gris-
rological and Communicative Disorders and
mother or the pregnancy that seem to Stroke. The following institutions participated: wold, who performed computations, and Peggy
Boston Lying-in Hospital; Brown University, Shaw and Bea Dean, who prepared the manu¬
be associated with a relatively high
Providence, RI; Charity Hospital, New Orleans; script.
risk of CP. If confirmed in other in¬ Children's Medical Center, Boston; Columbia Most of all, thanks are due to the parents and
vestigations, these newer observa- University, New York; Johns Hopkins Univer- children who participated in this study.
References
1. Myrianthopoulos NC, French KS: An appli- study of cerebral palsy in Western Australia role in brain,
in Grave GD (ed): Thyroid Hor-
cation of the US Bureau of the Census socioeco- 1956-1975: II. Spastic cerebral palsy and peri- mones and Brain Development. New York,
nomic index to a large, diversified patient popula- natal factors. Dev Med Child Neurol 1980;22: Raven Press, 1977, pp 73-91.
tion. Soc Sci Med 1968;2:283-299. 13-25. 14. Nunez J: Effects of thyroid hormones dur-
2. Broman S: The Collaborative Perinatal 8. Ellenberg JH, Nelson KB: Birthweight and ing brain differentiation. Mol Cell Endocrinol
Project: An overview, in Mednick SA, Harway M, gestational age in children with cerebral palsy or 1984;27:125-132.
Finello KM (eds): Handbook of Longitudinal seizure disorders. AJDC 1979;133:1044-1048. 15. Schlumpf M, Lichtensteiger W: Drugs and
Research. New York, Praeger Publishers, 1984, 9. Nelson KB, Ellenberg JH: Apgar scores as Hormones in Brain Development, Basel, Switz-
vol 2, pp 185-215. predictors of chronic neurologic disability. Pedi- erland, S Karger AG, 1982.
3. Nelson KB, Ellenberg JH: Obstetric com- atrics 1981;68:36-44. 16. Pfaff DW, McEwen BS: Actions of es-
plications as risk factors for cerebral palsy or 10. Myrianthopoulos NC, Chung CS: Congeni- trogens and progestins on nerve cells. Science
seizure disorders. JAMA 1984;251:1843-1848. tal malformations in singletons: Epidemiologic 1983;219:808-814.
4. Heinonen OP, Slone D, Shapiro S: Birth survey. Birth Defects 1974;10:1-58. 17. Hellman LM, Pritchard JA: Williams Ob-
Defects and Drugs in Pregnancy. Littleton, 11. Scott KG, Masi W: The outcome from and stetrics, ed 14, New York, Appleton-Century\x=req-\
Mass, Publishing Sciences Group Inc, 1977. utility of registers of risk, in Field TM (ed): Crofts, 1971, p 267.
5. Mantel N, Haenszel W: Statistical aspects Infants Born at Risk, New York, SP Medical & 18. Snavely SR, Hodges GR: The neurotox-
of the analysis of data from retrospective studies Scientific Books, 1979, pp 485-496. icity of antibacterial agents. Ann Intern Med
of disease. J Natl Cancer Inst 1959;22:719-748. 12. Philips JB, Dickman HM, Resnick MB, 1984;101:92-104.
6. Nelson KB, Ellenberg JH: Maternal seizure et al: Characteristics, mortality and outcome of 19. Alberman E, Stanley F: Guidelines for the
disorder, outcome of pregnancy, and neurologic higher-birthweight infants who require intensive epidemiologic approach, in Stanley F, Alberman
abnormalities in the children. Neurology 1982;32: care. Am J Obstet Gynecol 1984;149:875-879. E (eds): The Epidemiology of the Cerebral Pal-
1247-1254. 13. Sokoloff L: Biochemical mechanisms of the sies. Philadelphia, JB Lippincott Co, 1984, chap
7. Dale A, Stanley FJ: An epidemiological action of thyroid hormones: Relationship to their 14.

In Other AMA Journals


JAMA
Kawasaki Syndrome: Still a Mystery After 20 Years
D. M. Bell, MD (JAMA 1985;254:801)
Helping the Abused Child
D. A. Sargent, MD, JD (JAMA 1985;254:803)
The Accuracy of Experienced Physicians' Probability Estimates for Patients
With Sore Throats
R. M. Poses, MD; R. D. Cebul, MD; M. Collins, MD; S. S. Fager, MD (JAMA
1985;254:925-929)

Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 06/21/2015

You might also like