Objective To investigate changes in rates of cerebral palsy (CP) by birth weight, gestational age, severity of disability, clinical subtype and maternal age in the North of England, 1991–2000.
Methods Data on 908 cases of CP (816 singletons, 92 multiples) were analysed from the prospective population-based North of England Collaborative Cerebral Palsy Survey. Severity of disability, measured as a Lifestyle Assessment Score (LAS), was derived from the lifestyle assessment questionnaire. CP rates by birth weight, gestational age, birth weight standardised for gestational age and sex, severity of disability and maternal age were compared between 1991–1995 and 1996–2000 using rate ratios (RR).
Results The prevalence of CP in singletons was 2.46 (95% CI 2.29 to 2.63) per 1000 neonatal survivors compared to 11.06 per 1000 (95% CI 8.81 to 13.3) in multiples (RR 4.49, 95% CI 3.62 to 5.57), with no significant change between quinquennia. The singleton CP rates were higher for lower birth weight groups than birth weight ≥2500 g; and there were no significant changes for any birth weight group between quinquennia. There were also no changes in rates of more severe disability (LAS≥30%) by birth weight, gestation or clinical subtype. For preterm and term births the patterns of Z-score of birth weight-for-gestation are similar, with CP rates increasing as Z-score deviates from the optimal weight-for-gestation, which is about 1 SD above the mean.
Conclusions In contrast to increasing rates in previous years, rates of CP and more severe CP were stable by birth weight, gestational age and clinical subtype for 1991–2000.
Statistics from Altmetric.com
Cerebral palsy (CP) is the most common cause of significant motor impairment in childhood. Advances in antenatal and neonatal intensive care since the 1970s have resulted in improved survival of very preterm and very low birth weight (VLBW) infants, accompanied by an increase in CP rates in these groups.1,–,5 More recently a declining trend in CP rates in such small infants has been observed,5,–,7 and in some reports the decline is accompanied by a reduction in severity.8 Against this welcome trend, there has been no decrease in rates of CP in normal birth weight and term infants,7,–,9 who account for about a half of all children with CP.
The aim of this study is to investigate changes in CP rates in the North of England for 1991–2000 births in relation to birth weight, gestational age, severity of disability, clinical subtype and maternal age, using data from the prospective population-based survey of CP, the North of England Collaborative Cerebral Palsy Survey (NECCPS).
What is already known on this topic
▶ While the prevalence of cerebral palsy (CP) for term infants has been stable over the last three decades, rates of CP in preterm and low birth weight infants increased in the 1980s and 1990s; since then decreasing rates are reported by most studies.
▶ In North of England, rates of CP and more severe CP increased during 1964–1993, with the increases accounted for by low birth weight infants.
What this study adds
▶ In the North of England between 1991–1995 and 1996–2000, there was no increase in either overall CP rate in singletons, in rates by any birth weight group or in rates of more severe CP by birth weight, gestational age or clinical groups.
▶ For both preterm and term infants, the patterns of Z-score of birth weight-for-gestation are similar, with the CP rates increasing with deviations from the optimal weight for gestation at about 1 SD or more above the mean.
NECCPS uses the Little Club10 definition of cerebral palsy, updated by Bax.11 For this analysis, cases are those with congenital CP and include children who subsequently died. Cases of CP due to a known post-neonatal insult are excluded.
Study birth population
The Northern Region of England is a geographically defined area with a stable population of almost 3 million and approximately 31 000 births per year during the study period, residing in two main urban conurbations and extended rural areas. It comprises the counties of North Cumbria, Northumberland, Tyne and Wear, Durham and Darlington and Teesside. The number of registered births for the study period, 1991–2000, was obtained from the Office for National Statistics.
NECCPS prospectively records all infants with CP born to mothers resident in the region from 1991.12 Cases are notified to the survey by District Convenors, who are consultant community paediatricians. They coordinate services for such children and receive information about children needing services from other paediatricians, paediatric neurologists, physiotherapists, speech therapists and the regional child development centre. The convenor completes a notification form. Further details are forwarded to the survey when the child reaches 5 years of age to confirm the diagnosis and provide details of associated impairments. Further, parents are invited to return the lifestyle assessment questionnaire which records the impact of the CP on the child and family. It is very unusual for a case of CP to be diagnosed after age 6. However, the process of ascertainment by the convenor and the requirement to obtain parent consent means that sometimes children are added to the register up to age 8, even though diagnosed a year or two earlier. For the report up to year 2000 births, the dataset on which the analysis was based was locked down in 2009.
High case ascertainment is ensured because cases are notified from multiple sources, there is a regional network of interested clinicians and close links with the long standing prospective Perinatal Mortality Survey13 and Northern Congenital Abnormality Survey14 housed on the same premises. Every case of CP mentioned on a child death certificate and every case mentioned as co-morbidity on a late notification of a congenital abnormality is ascertained by the survey.
For this study we used anonymised data of children recorded on the NECCPS. The NECCPS has ethics permission, which includes permission to analyse anonymised data. Parents give permission for their child's name to be on the database and reports of analysis of anonymised data to be undertaken. Parents know what data are held about their child and contribute to the NECCPS data by completing a questionnaire.
Classification and definitions
CP is classified according to the agreement of the Surveillance of Cerebral Palsy in Europe:15 spastic CP (unilateral or bilateral), dyskinetic and ataxic.
Severity of CP was assessed using a measure of impact of disability on the child and family, the lifestyle assessment questionnaire. The questionnaire covers six dimensions: physical independence, mobility, clinical burden, schooling, economic burden and social integration, from which a global Lifestyle Assessment Score (LAS) is derived.16 17 The LAS is expressed as a percentage, with a maximally disadvantaged child scoring 100%. We defined children with LAS≤30% as a mild group, LAS=30–69% as moderate and LAS≥70% as severe. A typical child with a LAS of 30% completes most, but not all, self-help activities alone, poses little economic or social burden on the family and attends mainstream school with some assistance; whereas a typical child with a 70% score undertakes only a few self-help activities, requires specialised education and imposes a substantial economic and social burden on his family. Children who died before the LAS was measured are included in the most severe group (≥70%). Live cases with missing severity data were assigned a severity group according to the severity distribution in observed live born cases in the three birth weight groups (<1500 g, 1500–2499 g and ≥2500 g) and the five World Health Organization gestational age groups (<28, 28–31, 32–36, 37–41 and ≥42 weeks).
Severity of impairment was assessed using data on IQ and walking ability recorded at 5 years of age. The ‘severe’ group comprised children with IQ<50 or inability to walk without aids or both.
Preterm and term birth are defined as gestation less than 37 and 37–41 completed weeks respectively. Low birth weight (LBW) and VLBW are defined as <2500 g and <1500 g respectively.
We calculated CP prevalence as rates per 1000 neonatal survivors rather than live births. Denominator data for singleton neonatal survivors by birth weight groups were available for the Northern Region birth population for 1991–2000. As denominators by gestational age groups were not available, they were estimated from the gestational age distribution for Scottish singleton live births for the same years.18
An SD score (Z-score) of birth weight-for-gestation for each singleton case was derived from local standards by sex.19 Cases were allocated to Z-score bands chosen to equate to conventional percentiles (eg, in a normal distribution the Z-score band 1.28 to <1.88 is the same as 90th to <97th percentile). For calculation of CP rates by Z-score band, we estimated the number of neonatal survivors for each Z-score band assuming a normal distribution as recommended in previous reports.20 21
SPSS for Windows V.15.0 was used for the descriptive analyses. For comparison of rates between quinquennia, rate ratios (RR) and 95% CIs are presented with statistical significance at the 5% level. χ2 Test for differences in proportions and χ2 test for trend for rates over the 10-year period were used.
In all, 908 children with CP, not associated with a known post-neonatal insult, were recorded on NECCPS for 1991–2000 births; 816 singleton and 92 from multiple births (86 twins, 5 triplets and 1 quadruplet—10.1% of all children with CP). The CP rates for 1991–2000 were 2.46 (95% CI 2.29 to 2.63) per 1000 neonatal survivors in singletons and 11.06 per 1000 (95% CI 8.81 to 13.3) in multiples (RR 4.49, 95% CI 3.62 to 5.57).
Further analyses are restricted to singleton births. The CP rate was significantly higher in males (2.91 per 1000) than females (1.99 per 1000): RR 1.46 (95% CI 1.27 to 1.68). Figure 1 shows that rates of CP in singletons by year of birth and sex fluctuate over the 10-year period with no significant change in either direction (χ2 for trend=0.00, p=0.99). Over the same period there was a significant decline in singleton neonatal mortality from 4.8 to 3.6 per 1000 live births (χ2 for trend=16.38, p<0.0001). The proportion of deaths up to age 9 years from all singletons with CP decreased between quinquennia from 10.5% to 6.5% (χ2=4.06, p=0.04).
Table 1 shows that rates of CP in singletons did not change between quinquennia nor in any birth weight group (<1500 g, 1500–2499 g or ≥2500 g). The proportion of LBW infants among children with CP also remained stable at about 40%. There was also no change in the CP rates for children with severity ≥30% in any of the three birth weight groups. Table 1 also shows gestational age-specific rates of CP by quinquennia for all cases and for those with LAS≥30%. As with birth weight groups, there was, in general, no significant change in rates between quinquennia by gestational age group, although there was an increase for gestation ≥42 weeks. There was also no change in rate in any gestational age group for cases with LAS≥30%. The contribution of preterm births to all CP cases was about 40% in each quinquennium.
For children with the most severe CP (LAS≥70%), in 1996–2000 there was a significant decline in rates in the LBW infants from 5.50 to 3.43 per 1000 (RR 0.63, 95% CI 0.41 to 0.98), but no significant change in infants ≥2500 g. In preterm infants there was a similar decline in rates for children with the LAS≥70% from 5.09 to 3.97 but it did not reach statistical significance; and there was no significant change in term infants.
Table 2 shows that term births accounted for most non-spastic CP cases (72%). Among spastic CP cases, about half of bilateral spastic cases were preterm compared to about 30% among the unilateral type. There was a higher proportion of more severe cases (LAS≥30%) in children with bilateral spastic CP and non-spastic CP than in children with unilateral spastic CP. Despite a decline in the proportion of more severe cases (LAS≥30%) in children with spastic CP from 86.9% to 80.6% between quinquennia (χ2=5.96, p=0.01), the rate of more severe spastic CP did not change significantly. Table 2 also shows severity based on measures of intrinsic impairment. The proportion of cases with severe impairment (measured as IQ<50 or inability to walk without aids or both) did not reduce significantly in either children with spastic CP (χ2=0.78, p=0.38) or children with all clinical subtypes (χ2=1.14, p=0.28). All clinical subtypes of CP were more common in infants with birth weight-for-gestation <50th percentile and about a third were <25th percentile (table 2).
For term and preterm infants, the CP rate is the lowest at 1 SD above the average birth weight for gestation, increasing with deviations in either direction from this optimum (figure 2). The reversed J-shaped relationship for preterm births was more distinct when a fetal growth standard was used which was calculated from the mean term birth weights from the local standard19 applying the Gardosi formula22 (data not shown). Figure 3 shows that CP rates by Z-score band were higher in males.
Maternal age-specific CP rates were significantly higher for younger mothers, in particular for those <20 years, compared with the reference group of mothers aged 25–29 years with the lowest rate (table 3). Children of older mothers (≥35 years) had a 20% higher CP rate than those of the reference group but this did not reach statistical significance.
This population-based study in the North of England found no temporal changes in either overall or any birth weight specific CP rates for 1991–2000 births. There were also no changes in rates of more severe CP by birth weight, gestational age or clinical subtype. This is important because one can be more confident of complete ascertainment of children with more severe CP.3 16 Our study confirmed findings of a larger CP cohort21 that, for term and preterm infants, the CP rate is lowest at 1 SD above the average birth weight-for-gestation, increasing with deviations in either direction from this optimum. Males had higher CP rates in all birth weight-for-gestation Z-score bands but the difference was only significant in the two higher bands.
These results are encouraging in view of previous studies based on data from three districts of the same region which reported increasing rates of CP for all LBW infants and for more severe cases (LAS≥30%) in 1961–1993 births.3 As the current study covered a larger area than the two previous North of England studies,3 23 we undertook a sub-analysis of data from the three districts only (which account for about one fifth of the CP cases) and found that, similar to the regional trends, there was no significant change in CP rates in those districts between quinquennia. For 1984–1993 LBW infants accounted for 49% of cases of CP compared with 34% for 1964–1973 births.3 In the current study, the proportion of LBW infants was 40% and did not change significantly over the study period. The previously reported trends23 for CP rates for preterm singleton infants and for the contribution of preterm infants to all CP cases to increase, were also arrested, despite the fact that rates of preterm birth have increased.24 25 The stabilisation of CP rates found in our study for 1991–2000 and the evidence of a lower contribution of LBW infants in children with the most severe CP (LAS≥70) may indicate further improvement in perinatal care. It is consistent with other studies reporting reduced neonatal morbidity and CP in extremely LBW or VLBW children born 1990s to early 2000s.26 27 Wilson-Costello et al27 explain this improvement by further advances in perinatal care, including increased antenatal steroid use and caesarean section delivery, better management of sepsis and decreased postnatal steroid use; changes in practice which occurred in Northern England during the period we report.
There are advantages in reporting from a single large register rather than from pooled data.26 28 Our data were collected prospectively from a population in a well-defined geographical area. Consistent inclusion criteria were used and we did not have to restrict analyses to variables collected by other registers; for example, we are the only register in the world to capture the impact of CP on child and family. Nor did we have to deal with missing data for some years from some registers, or inadequate denominator data. While a register can never be certain it has 100% ascertainment, at least in a single register the same notification procedures are used, in our case using a well-established network of local clinicians. However our study has some limitations. Live cases with missing severity data (about 20%) were assigned to a severity group as described in the Methods section. It is reassuring that the severity distribution was almost identical whether cases were allocated by birth weight or clinical subtype. Also, as described in the Methods section, we had to use a Scottish dataset for denominators by gestational age; and assume a normal distribution of neonatal survivors by Z-score band.
As in our study, Northern Ireland found no significant change in rate of CP for 1981–1997 births; however our rates of CP for VLBW infants (<1500 g) were higher at about 75 per 1000 than Northern Ireland's at 44.5 per 1000.8 This may be partially explained by our use of neonatal survivors rather than live births as denominators. This can substantially increase the CP rates in VLBW infants due to their high neonatal mortality; for example, our rates decrease to 58 per 1000 for 1991–2000 when calculated per 1000 live births. A UK study, using pooled data from five registers, including NECCPS, assessed trends in CP rates by birth weight and severity of motor impairment for 1976–1999.28 Despite being at a greater risk of CP, lighter infants were less likely to develop the most severe forms of motor impairment than normal birth weight infants. In our study, which used a different measure of severity, the proportion of most severe cases of CP (LAS≥70%) in the second half of the study period was about 60% lower in VLBW infants than in those weighing ≥2500 g, but it was not statistically significant (p=0.06). The findings of our study were consistent with other studies for normal birth weight and term infants, showing a stable prevalence of CP during the 1990s.7 8 For moderately LBW infants (1500–2499 g), as in other studies,7 our CP rates declined in all severity groups but the trends were not statistically significant.
CP rates were over 40% higher for infants born to mothers under 20 years of age compared to the reference group of 25–29-year-olds which had the lowest CP rates. This mirrors the excess neonatal mortality for mothers under 20 over the same period in the North of England.29 Other studies found advanced maternal age (≥35 years) an independent risk factor for CP in term infants30 and in infants of all gestational ages combined.31 Although we found a 20% increase in CP rates for infants of mothers aged ≥35 years, it did not reach statistical significance, even after restricting to term births.
The important message from this study is that, in contrast to previous years, there were no increases in CP rates in any birth weight or gestational age group; neither did rates of more severe cases (which are more reliably ascertained) increase, whether assessed as impact of severity of disability on the family or by assessment of intrinsic impairment (IQ and walking ability). Unfortunately, CP rates did not fall despite the decline in neonatal mortality. An encouraging finding was that for children with the most severe CP (LAS≥70%), there was a significant decline in CP rates for 1996–2000 in the LBW group of infants, although there was no decline for infants ≥2500 g.
We thank the parents of the children with cerebral palsy for their involvement and interest in the survey; and Mary Bythell at the Northern Region Maternity Survey Office for administrative support. We pay tribute to Dr Edmund Hey who died at the end of 2009. Edmund's idea was the inspiration for the register, and he then encouraged Professor Stephen Jarvis to develop it.
The collaborators during the period of data collection were: R Carpenter and R Willoughby, Darlington; S Pandy and N Cookey, Durham; A Paynter and P Whitehead, East Cumbria; R Menzies, Gateshead; K Agrawal, Hartlepool; M Gibson and S Wariyar, Newcastle; C Jessen, Northumberland; B Kurup, North Tees; E Lee, South Tees; H Palmer and N Brewster, South Tyneside; A Johnson, South West Durham; K Horridge, Sunderland; S Precious, S Bilsberry and V Wood, West Cumbria
Funding JR is funded by a Personal Award Scheme Career Scientist Award from the National Institute of Health Research (UK Department of Health). Administrative support for the NECCPS, as part of the Northern Regional Maternity Survey Office, is funded by the Regional Health Authority, District Health Authorities and Primary Care Trusts.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.