Cerebral palsy and restricted growth status at birth: population-based case–control study


Prof J Gardosi, Perinatal Institute, Crystal Court, Aston Cross, Birmingham B6 5RQ, UK. Email jason.gardosi@pi.nhs.uk


Objective  To evaluate the association between growth status at birth and subsequent development of cerebral palsy in preterm and term infants.

Design  Population-based case–controlled study.

Setting  Cerebral palsy register in Western Sweden.

Subjects  Cohort of 334 singletons born between 1983 and 1990, with cerebral palsy diagnosed from age 4, and 668 singletons matched for gestation, gender and delivery unit.

Method  Growth status at birth was determined using small for gestational age (SGA) categories, with customised birthweight percentiles (SGAcust) based on the Swedish population.

Main outcome measures  Proportion of babies that were SGAcust, comparing cases and controls in three gestational age categories: early preterm (24–33 weeks), late preterm (34–36 weeks) and term (37+ weeks).

Results  Of the 334 children with cerebral palsy, 87 (26.6%) were born early preterm, 27 (8.1%) late preterm and 218 (66%) at term. Children who had been born at term were more likely to have been SGA <1st customised percentile (SGAcust1) than their matched controls (OR 6.6, 95% CI 2.3–18.6). In contrast, children with cerebral palsy born preterm were not more likely to have been SGAcust1 (OR 0.9, 95% CI 0.4–1.9), and this applied to early preterm as well as late preterm births. For less severely small babies (SGA between 1st and 5th customised percentiles), the association with cerebral palsy remained significant for term births (OR 5.2, 95% CI 2.7–10.1) but was again not significant for preterm births.

Conclusions  Term singletons with severely SGA birthweights had a five- to seven-fold risk of developing cerebral palsy compared with gestational age-matched infants with birthweights within normal limits. For children born preterm, SGA was not more likely to be present in cases than in controls. These findings support the concept of cerebral palsy as a multifactorial condition and highlight the importance of antenatal surveillance of fetal growth.


Cerebral palsy is a syndrome complex characterised by an aberrant control of movement or posture, which appears early in life and leads to lifelong motor disability. It is the most common physical disability in childhood, present in around 2 per 1000 live births.1 While perinatal mortality has decreased over recent decades, there has been no overall change in prevalence in industrialised countries,1,2 and the underlying causes are still poorly understood.

An association between cerebral palsy and intrauterine growth restriction has been suspected for some time3 but is difficult to prove as information about antenatal growth is usually not available in cohort studies large enough to assess cerebral palsy as an outcome. Instead, low weight at birth referenced to population norms is generally used as a proxy for growth restriction. However, this is imprecise as it includes not only pathological growth restriction but also constitutional smallness. This may be a reason why the relationship between cerebral palsy and fetal growth restriction is not well defined, and why it has even been described as paradoxical.4 There is also uncertainty about the strength of the link between cerebral palsy and growth restriction at term versus preterm gestations.5

The actual growth status at birth can now be assessed using customised birthweight percentiles based on the growth potential calculated for each baby.6 This method has allowed a distinction between pathological and constitutional smallness and found that population-based birthweight standards result in a substantial number of false-positive and false-negative assessments of ‘smallness’. Babies who are small for gestational age (SGA) by customised percentiles are significantly more likely to have abnormal umbilical artery Doppler flow, caesarean section for fetal distress, low Apgar scores, admission to special neonatal care and prolonged hospital stay, as well as stillbirth and neonatal death than babies who are SGA by the population based standard. In fact, babies who are SGA by population weight percentiles only are similar to those with birthweights within normal limits.6–9 We therefore applied this method to a large cohort of children with cerebral palsy to study the relationship between cerebral palsy, fetal maturity and growth status at birth.


Approval for the study was obtained from the ethics committee in Göteborg. Two investigators, unaware of the paediatric outcome, scrutinised all records documenting variables of interest.


The internationally accepted Swedish classification of cerebral palsy and its subtypes was used, which defines cerebral palsy as a group of nonprogressive but often changing motor impairment syndromes, secondary to lesions or abnormalities of the brain arising in the early stages of development.10 The terminology was updated according to recommendations of the network for Surveillance of Cerebral Palsy in Europe (www-rheop.ujf-grenoble.fr/scpe2/site_scpe/). All children included were at least 4 years old at time of diagnosis.11


The data were derived from the 1983–90 birth cohort of the cerebral palsy project in Western Sweden, which has been previously described.11 Cases were included if the child was born in Sweden, had lived in the study area at 4–8 years of life during the study period and had no obvious postnatal cause of cerebral palsy.


There were a total of 397 cases on the register during the study period. Of these, 63 were excluded because of multifetal pregnancy (n = 33), because cerebral palsy was considered to have had postnatal origins (n = 21) or because the data were incomplete (n = 9). This left a total of 334 cases, which included diagnoses of bilateral spastic cerebral palsy (n = 150, 44.9%), unilateral spastic cerebral palsy (n = 122, 36.5%), tetraplegia (n = 25, 7.5%), ataxia (n = 17, 5.1%) and dyskinetic syndromes (n = 20, 6.0%). Because of small numbers, subgroup analysis was undertaken in the two largest diagnostic categories only.


Every case was matched with two controls from the Swedish National Birth Register. The closest births occurring before and after the index birth were chosen. Controls were matched for gestational age, gender and delivery ward. Matching for gestational age, sex and multiple pregnancies was complete in all cases, and matching with regard to delivery ward was complete in 93%, as controls could not always be recruited from small units. In these instances, controls were recruited from another unit of similar level and size. A total of 668 matched controls were thus selected.

Statistical methods

Gestational age was based on ultrasound scan in 97% of pregnancies and was usually performed between 16 and 19 weeks of gestation. The remaining 3% of pregnancies were dated by last menstrual period. Three gestational age categories are defined in Table 1.

Table 1.  Glossary of terms and definition of cut-offs used
  • *

    Calculated according to early pregnancy ultrasound scan.

  • **

    Birthweight for gestation percentiles, adjusted for maternal height, weight in early pregnancy, parity, ethnic origin and baby’s sex.

Gestational age at birth*Weeks, days
Preterm (all)<37.0
Early preterm<34.0
Late preterm34.0–36.6
Growth status a birth**Customised percentile
Not restricted≥10
Growth restricted<10
Severely growth restricted<1

Cases and controls were compared by categorical variables (smoking, alcohol, baby gender and parity) and continuous variables (birthweight, gestational length, maternal height and weight). Maternal weight and birthweight were tested for normality using the Shapiro–Francia (W) test. T test was used for birthweight and Mann–Whitney test for maternal weight to compare averages between cases and controls.

Growth status was determined using a customised birthweight-for-gestation standard according to methods described previously.6 Using coefficients derived from a Swedish population,8 adjustments were made for all known constitutional variables affecting fetal growth: maternal height, maternal weight in early pregnancy, parity, ethnic origin and sex. These were combined with a proportionality curve6 based on longitudinally derived Swedish fetal weight curves for male and female infants.12

Customised birthweight percentiles were grouped into categories (SGA <1, <10 centiles) to assess various degrees of restricted growth status at birth (Table 1). Comparative analysis was undertaken using odds ratios (OR) and 95% confidence intervals (CI), with comparisons based on the 10–90 customised percentile range. Analysis was undertaken in three gestational age categories: early preterm (<34 weeks), late preterm (34 weeks 0 days to 36 weeks 6 days) and term (37 weeks 0 days and over).


One hundred and sixteen of the 334 children with cerebral palsy (34.7%) were born preterm (<37 weeks of gestation), including 89 (26.6%) born before 34 weeks and 27 (8.1%) between 34 and 36 weeks.

Table 2 shows that cases and controls were similar in maternal height, maternal weight, parity and sex ratio. Mothers of children who developed cerebral palsy were more likely to have smoked in pregnancy than mothers of controls (37.5 versus 28.3%, P < 0.01). They were also more likely to have drunk alcohol during pregnancy, although the difference did not reach statistical significance (49.7 versus 44.9%, P = 0.16).

Table 2.  Univariate analysis of cerebral palsy cases versus controls
 Cases (n = 334)Controls (n = 668)P
  1. NS, not significant.

Maternal weight (kg)
Maternal height* (cm)
Parity, n (%)
Primipara152 (45.5)302 (45.2)NS
Baby’s sex
Male, n (%)179 (53.6)358 (53.6)NS
Male/female ratio1.151.15 
Smoking, n (%)
None194 (58.1)447 (66.9) 
Any125 (37.5)189 (28.3)<0.01
Unknown15 (4.5)32 (4.8) 
Alcohol, n (%)
None149 (44.6)327 (49.0)NS
Any166 (49.7)300 (44.9) 
Unknown19 (5.7)41 (6.1) 

Table 3 shows the comparison of average gestational age and birthweight for each of the gestational age categories. Length of pregnancy was similar for cases and controls as a result of the method used to match controls for gestation. Cases had a significantly lower median birthweight if they were born at term but were similar to controls in both of the two preterm categories.

Table 3.  Comparison of cases and controls in three gestational age groups
 Early preterm (<34 weeks)Late preterm (34–36 weeks)Term (37+ weeks)
  1. NS, not significant.

Number89178 2754 218436 
Gestation (median, days)2122120.862462470.65281281NS
Birthweight (median, g)148514600.57243025700.2633753590<0.01

Analysis of fetal growth status by customised percentiles is presented in Table 4. Children with cerebral palsy who were born at term were significantly more likely to have had a restricted growth status at birth than their controls (<1st centile: OR 6.6, 95% CI 2.3–18.6; 1st–5th centile: OR 5.2, 95% CI 2.7–10.1). Looking at the main diagnostic subcategories, the association was significant for unilateral spastic cerebral palsy (n = 98: OR 6.7, 95% CI 1.7–26.2) and was also suggested but did not reach significance for bilateral spastic cerebral palsy (n = 66: OR 4.7, 95% CI 0.8–26.8).

Table 4.  Comparison of cases versus controls at preterm and term, for categories of SGA according to customised percentiles
Gestational age at birthCustomised centileCasesControlsOR95% CI
n (%)n (%)
  • *

    Odds ratios are referenced to the 10–90 centile group of each gestation category.

Preterm (<37 weeks) n = 116n = 232 
<111 (9.5)24 (10.3)0.90.4–1.9
1–58 (6.9)27 (11.6)0.60.2–1.3
<1027 (23.3)60 (25.9)0.90.5–1.4
10–9083 (71.6)158 (68.1)* 
>906 (5.2)14 (6.0)0.80.3–2.2
Early preterm (<34 weeks) n = 89n = 178 
<1020 (22.5)47 (26.4)0.80.4–1.4
10–9066 (74.2)122 (68.5)* 
Late preterm (34–36 weeks) n =27n =54 
<107 (25.9)13 (24.1)1.10.4–3.4
10–9017 (63.0)36 (66.7)* 
Term n =218n =436 
<114 (6.4)5 (1.1)6.62.3–18.6
1–531 (14.2)14 (3.2)5.22.7–10.1
<1054 (24.8)42 (9.6)3.01.9–4.7
10–90152 (69.7)358 (82.1)* 
>9012 (5.5)36 (8.3)0.80.4–1.6

In contrast, preterm children who developed cerebral palsy did not have an increased rate of restricted growth status (SGAcust <1st centile) compared with their gestational age-matched controls. As shown in Table 4, this applied for early preterm (<34 weeks: OR 0.8, 95% CI 0.4–1.4) as well as late preterm babies (34–36 weeks: OR 1.1, 95% CI 0.4–3.4). Furthermore, there was no significant difference for preterm babies in either of the main diagnostic subgroups. A similar result was obtained when the gestational age cut-off for ‘early preterm’ was set at <32 instead of <34 weeks.


To our knowledge, this is the first study to investigate the association between the development of cerebral palsy and fetal growth status at birth, as assessed by customised birthweight percentiles. This is possible because the database includes information not only about precise gestation length and weight at birth but also about the constitutional variables known to affect fetal growth. In most instances, the actual course of intrauterine growth is not known, but diminished growth status at birth can be inferred from a weight which is below an individually adjusted range of normal.

We show that the risk of developing cerebral palsy is linked to the severity of restricted growth status at birth and that this is only the case if the pregnancy had reached term. In babies born preterm, there was no difference between cases and controls.

These findings are in part similar, and in part contrasting to previous studies, which used unadjusted, population-based norms to determine relative size at birth. An association between SGA at term and cerebral palsy has been demonstrated.5,11,13 However, for babies born preterm, the relationship has so far been less clear, with some studies suggesting an association with SGA at 34–36 weeks,13–15 and others an association at gestations <32 weeks only.16–18 Our results found no increased risk of cerebral palsy for babies with restricted growth status who were born at early or late preterm gestations.

Prematurity, spontaneous as well as iatrogenic, is itself known to be associated with fetal growth restriction.19 This is also suggested in the current study as 23.3% of preterm cases and 25.9% of preterm controls had birthweights below the 10th customised percentile (Table 4). It is therefore appropriate to use fetal rather than neonatal weight standards in the study of cerebral palsy.14 However, prematurity is also associated with maternal height and weight,20,21 and both of which are also in themselves factors which affect birthweight.6 These confounders may have affected results in other studies, while they are here adjusted for by individually customised percentiles based on an ‘optimal’ intrauterine weight curve for each pregnancy.

Jarvis et al.14 found an increased risk of cerebral palsy when birthweight was large for gestational age, which was not observed in the current analysis. Although theirs was a larger study, it assessed birthweight against general population-based norms, rather than a matched control group. Furthermore, our method was here again able to distinguish between constitutional and pathological variation in birthweight. As shown in diabetic pregnancies, the use of customised percentiles for defining ‘large for gestational age’ is better able to identify pathological macrosomia than population-based percentiles.22

The application of customised percentiles has highlighted stronger associations between fetal growth status and a number of perinatal indicators of adverse outcome, including stillbirth and neonatal death,8 and antenatal and postnatal morbidity, including fetal distress and prolonged stay in special care.9,23 In the current study, the same method for assessing growth status at birth illustrates significant links with the subsequent development of cerebral palsy.

Most instances of fetal growth restriction are associated with placental insufficiency. Some specific changes have been observed in the brain of the growth-restricted infant, including altered cerebral haemodynamics, reduced grey matter volume24 and reduced total DNA in glial cells and neurons.25 This is also supported by animal studies showing reduced oxygen delivery to the brain and restricted growth of the forebrain and cerebellum.26,27 However, cerebral palsy following preterm birth may be associated with the immaturity of brain development itself.28

It is tempting to hypothesise that a prolonged period of reduced intrauterine nutrition will put the baby at increased risk of developing cerebral palsy. This notion would be consistent with evidence from magnetic resonance imaging studies, which suggests that approximately 75% of brain lesions associated with cerebral palsy occurs in the early or middle part of the third trimester.28,29 While early delivery may lead to neonatal and delayed complications associated with prematurity, including cerebral palsy, spontaneous preterm labour following intrauterine growth restriction could in many instances be a fetal adaptive response, an ‘escape’ from an unfavourable intrauterine environment.19 Our findings confirm the link between fetal growth restriction and preterm birth, but further work is needed to establish why some of these premature babies proceed to develop cerebral palsy while others do not. However, most instances of cerebral palsy occur after birth at term, as was the case in two-thirds of our study population, and a prolonged period of poor intrauterine nutrition leading to restricted growth status at birth appears to be associated with an increased risk of developing cerebral palsy.

These findings are relevant for considerations of prospective clinical management and prevention. They put the onus on the clinical service to improve on the antenatal recognition of intrauterine growth restriction, to be able to better advise mothers if there is risk of adverse outcome for the baby, including an increased likelihood of developing cerebral palsy.

Contribution to authorship

B.J. and J.G. conceived and designed the study, analysed and interpreted the data, wrote the manuscript and made revisions. A.F. carried out statistical analysis and interpreted the results. K.A., G.H. and H.H. assisted with maintenance of cerebral palsy register, interpretation of results and revisions of manuscript.

Conflict of interest

None of the authors had a conflict of interest relating to this study.

Ethics approval

The study was approved by the Goteborg Ethics Committee.