Back

Chief Scientist Office - Final Report

1. Summary

The Scottish Register of Children with a Motor Deficit of Central Origin was established by the Public Health Research Unit in Glasgow in 1990 and contains information on cases of children with a motor deficit of central origin born from 1984 onwards. The administration of the register has recently been taken over by the Information and Statistics Division (ISD) of the National Health Service in Scotland. In order to investigate factors predictive of cerebral palsy (CP), the CP cases from the register were linked to the Scottish Morbidity Record for maternity (SMR2) data.

This linkage gave access to additional information on the CP cases and allowed the SMR2 data to be used to calculate risks of CP for each factor. The factors that were found to be predictors of CP when considered on their own were:

Problems with the umbilical cord were not found to be predictive of CP. Neither was a previous history of spontaneous abortions or a history of previous Caesarian sections.

It was ascertained that the increased risk in twins could not be explained by the fact that twins are, on average born lighter and earlier. In fact the risk of CP at early gestations for singletons is higher than the risk for twins. It therefore appears that being a twin is a risk in itself. It was also shown that when one twin is stillborn or dies shortly after birth the surviving twin has an increased risk of CP.

The results suggest that the aetiology of CP is different in pre-term and term babies. In the former the lower the birthweight and the earlier the gestational age the higher the risk of CP. For this group the occurrence of pre-eclampsia was also found to be a significantly protective. For term babies the mode of delivery was found to be important with higher rates of CP among forceps, breech and emergency Caesarian sections and lower rates among elective Caesarian sections. Spastic bilateral CP occurs more commonly among babies of low birthweight and early gestational age. Spastic hemiplegia CP is more common at later gestations.

Back to top

2. Original aims

The aim of this project was to construct a retrospective cohort study to examine the relationship between various antenatal and perinatal factors and the subsequent risk of CP. This was to be done by using record linkage techniques to join the files from children on the Scottish Register of Motor Deficit of Central Origin to their mothers SMR2 maternity files for the years 1984 to 1990. SMR2 data for all children born in these years could then be examined, and the risks of CP for each factor calculated, with statistical adjustments where necessary.

As well as looking at predictive factors of CP overall, the aim was to investigate possible differences in aetiology between single and multiple births and between pre-term and term babies. A further objective was to determine whether or not different types of CP had different origins.

A secondary aim was to ascertain the quality of the data held on the register by comparing those variables common to both the register and the SMR2 forms.

Back to top

3. Methods

The CP data were linked to the SMR2 delivery records by ISD. This was carried out on a one-pass best-link basis whereby each CP record was allowed to make a best possible match to a corresponding SMR2 record. The identifying items used were surname soundex of child compared to the surname or maiden name soundex of mother, childs date of birth, mothers date of birth, postcode of residence at the time of birth and childs sex.

Once the linkage was achieved the quality of the data on the register was checked by comparing the information common to both registers. Where discrepancies were found the SMR2 record was assumed to be the more accurate.

The information available from linking with the SMR2 data can be divided into 4 main categories:

The data were analysed using logistic regression. A preliminary analysis focussed on the data common to both registers and used both the cases in which a link to an SMR2 record was established and those where no link was found.

More detailed analyses were carried out using the information only available for the linked data. Particular attention was paid to comparing the results for pre-term and term babies and the analysis was also repeated for different types of CP.

Back to top

4. Results

Numbers of children included in the study

Between 1984 and 1990 there were 554,147 total births in Scotland. The number of registered cases of CP in children born in these years was 839. Of these, 31 cases were identified as having postnatal CP and were therefore excluded from the analysis. Also omitted were cases of CP with a recognised syndrome of which CP is a known manifestation, those who were diagnosed before 2 years old and those children whose mothers were not resident in Scotland at the time of the birth. This left 646 cases of which 586 were singleton births and 60 were twin births.

Record Linkage Accuracy and Completeness

Of the 646 cases of CP described above, 570 (88.2%) were successfully linked to the delivery records. The failure rate is higher than usual for this type of linkage probably due to the linkage being between the record of the child and the record of the mother. A comparison between the cases that linked and those that did not was carried out and no significant differences between the two groups were found with respect to the variables common to both data sources.

The common variables are information on birthweight, gestational age, age of the mother and data on previous pregnancies. The extent of the agreement in birthweight and gestational age between the two records was found to be line with results from previous studies into the accuracy of maternal recall and the differences were not found to have any significant effect on the results. A few, though not all, of the more extreme discrepancies were attributable to clerical errors in the register.

Differences in some of the other variables common to both the register and the SMR2 forms were found to be more substantial. It appears that data such as number of previous pregnancies are not accurately recorded on the register.

Risk factor analysis

Preliminary analysis using both linked and unlinked cases confirmed previous findings that the rate of CP is substantially higher when birthweight and/or gestational age are low. It was also confirmed that the odds of having CP are higher for males than for females, the relative risk for males being 1.24 (1.07,1.48) compared to females.

The relative risk for twins compared to singletons is 4.83 (3.80,6.13). However, for the smallest babies the rate of CP in singletons was found to be higher than the corresponding rate for twins. At higher birthweights this is reversed. Overall the risk of CP for twins is more evenly spread over the range of birthweights. The same is true for gestational age.

From the data linkage it is possible to ascertain the outcome for the co-twin in those cases that linked successfully. The relative risk in cases where one twin was either stillborn or died neonatally of the co-twin having CP is 6.3 (3.1,12.8) compared to cases where both twins survive. The birth order was not found to be significant.

An analysis was then carried out using the linked data for singleton births only. The factors that were found to be associated with CP when considered on their own were:

No association was found between cord problems and rate of CP. Problems with the umbilical cord encompassed prolapse of the umbilical cord, umbilical cord around neck and umbilical cord entanglement.

When the factors were considered together the only variables that remained significant were birthweight, gestational age, age of the mother, Apgar score, sex of the baby and mode of delivery.

The analysis was then repeated for preterm (<37 weeks) and term babies (37 weeks or more). It was found that in preterm births birthweight and gestational age were still highly significant. Once these were adjusted for the only variable which remained significant was pre-eclampsia. Rates of CP were found to be lower when the mother suffered from pre-eclampsia. Apgar score was not a significant predictor. Neither was sex of the baby.

The significant factors for the term births were substantially different and were more concerned with the birth itself. Birthweight, Apgar score, mode of delivery and maternal age remained significant when other factors were taken into account.

Four types of CP were recorded on the register spastic hemiplegia, spastic bilateral, dyskinetic and ataxic CP. The number of cases for the latter two types was small and therefore the comparison by type has focussed on differences between spastic hemiplegia and spastic bilateral. A higher percentage of the spastic bilateral cases were found to be in low birthweight/early gestation births compared to the spastic hemiplegia. When the effect of gestational age was taken into account the risk of spastic bilateral CP was much more dependant on birthweight than the risk of spastic hemiplegia CP.

Back to top

5. Discussion

A recurring problem in the analysis of CP data is the lack of reliable denominator data. In particular many previous studies have been restricted to either preterm or term infants because of lack of access to data on gestational age. The ability to link the data from the CP register with data from the SMR2 forms gives access to both good denominator data and to reliable information on gestational age.

The present study confirmed that the rate of CP is higher in twins than in singletons. Since twins are born earlier and lighter on average this might be expected. For both groups the highest rates of CP are at the earliest gestational ages and lowest birthweights. However at the lowest gestational ages/birthweights the rate of CP for twins is below the rate for singletons.

At the same gestational age, twin babies will be lighter than singleton babies will and it is therefore surprising that there is such a clear indication of a lower rate for twins at these early gestations. It must be borne in mind that a birth at an early gestation is a more normal outcome for a twin pregnancy whereas such a birth is unusual in a singleton pregnancy and may indicate some fetal insult. Overall the rates for twins vary less with birthweight and gestational age than the rates for singletons and therefore the observed higher risk of CP among twins cannot simply be attributed to their lower birthweights and earlier gestational ages. Being a twin of itself increases the risk of CP.

It was found that the rates are higher for low birthweight and for pre-term birth. However for babies born before 32 weeks gestation the rate of CP is lower in small for gestational age babies than it is in those with birthweights above the 10th centile. This result is most likely explained by differences in the rates of stillbirths and neonatal deaths that are greater in low birthweight/pre-term babies. In particular a higher rate of CP might have been observed among those who were stillborn or died within the first months of life, had they survived long enough to have the condition diagnosed.

Overall a large number of the factors included in the study appeared to be predictive of CP when examined individually but were not found to be significant when the effect of gestational age and birthweight was adjusted for. Many of these factors have recently been the subject of great interest. For example, maternal infection has been suggested as an important factor. These data suggest that there appears to be a higher rate of CP when a maternal infection was recorded. However, infection may lead to pre-term delivery and therefore it may be that the apparent effect of this condition is just a marker for early gestational age.

Previous studies have looked at the relationship between pre-eclampsia and CP and have concluded that the occurrence of pre-eclampsia is protective of CP. This is borne out by the Scottish data, where the rate of CP is lower in those mothers who suffered from pre-eclampsia than in those who did not. This effect is significant among pre-term babies. There is no evidence that pre-eclampsia has any effect on the risk of CP among term babies.

The mode of delivery was found to be an important factor for term babies but was not significant for pre-term babies. In the former a higher rate of CP was found among breech and forceps deliveries. The rate was lower when an elective Caesarian section was performed. For emergency Caesarian section the rate of CP was lower if no labour occurred and higher if the section was performed after the onset of labour. This suggests that it may be the problem that led to an emergency Caesarian section being performed that is the cause of the CP rather than the section itself.

Deprivation category was of borderline significance as a factor in most of the models. The general pattern was a higher rate of CP in the more affluent areas and a lower rate in the most deprived. Given that both birthweight and gestational age decrease, on average, as the level of deprivation increases this is a surprising result. One possible explanation is that the level of registration may be higher in more affluent areas and that therefore the difference is due to under reporting in the more deprived areas.

It might be expected that children with CP would have a low Apgar score at birth. In general this was found to be true but Apgar score was not found to be a predictor of CP in pre-term babies. It was observed that Apgar score was poorly recorded in pre-term babies, and its interpretation is thought to be complex.

Back to top

6. Conclusions

The fact the twins are born earlier and lighter cannot explain the higher rate of CP in twins on average than singletons. When one twin is stillborn or dies neonatally, the risk of CP in the surviving twin in increased.

The aetiology of CP in preterm and term births is different. Pre-term CP is linked to birthweight, gestational age and pre-eclampsia. Term CP is linked to birthweight, maternal age and mode of delivery.

The aetiology of spastic bilateral and spastic hemiplegia is also different, the former occurs more frequently in low birthweight/ early gestational age births.

Back to top

7. Importance to NHS

The ability to set up this sort of cohort study at a national level exists only in Scotland and some of the Scandinavian countries. CP is an important cause of lifelong disability and handicap, yet its aetiology is poorly understood. Studies such as this, by looking at risk factors within the whole population and controlling for confounding variables, contribute to our understanding of aetiology, and hence point to methods of prevention.

CP is also an important cause of litigation against the NHS. At one time, it was considered that the majority of cases were the result of mismanagement during delivery. This study provides further evidence that for the majority of cases, the cause occurs long before the delivery.

Back to top

8. Future Research

From the linked data used in the project the main focus was on singleton births. Further research into the factors predictive of CP in twins could be informative since it is clear that there are differences in the pattern of occurrence in twins and singletons.

It is well established that low birthweight and low gestational age are both high risk factors for CP. Rates of still births and neonatal deaths are also going to be higher in these groups however and there is a need to carry out an analysis of the data that takes into account the rate of still births and neonatal deaths.

There is a need to ensure that the registration of cases onto the register continues and includes as many cases as possible. The present study could only look at births occurring between 1984 and 1990. Keeping the register up to date would enable trends in the occurrence of CP to be investigated. This would also allow the effects of changes in obstetric practices to be studied.

Back to top

Cerebral Palsy Register for Scotland
Edinburgh Napier University, Merchiston Campus
Edinburgh EH10 5DT

cprs@napier.ac.uk