This article is commented on by Himmelmann on page 876 of this issue.
Ms Susan Reid at Developmental Medicine, Royal Children’s Hospital, Flemington Road, Parkville, Victoria 3052, Australia. E-mail: email@example.com
Aim The aim of this study was to assess overall and gestational age-specific trends in the rate of cerebral palsy (CP) in Victoria, Australia, and to compare these findings with other population data.
Method Individuals born in Victoria from 1970 to 2004 with non-postneonatally acquired CP were identified from a population register; 3491 were included in the study (1963 males, 1528 females). After a literature review, comparison data were extracted from publications using previously devised inclusion criteria. Rates were calculated per 1000 live births for all CP and by gestational age group: these were tabulated and plotted by year of birth.
Results Data from nine registries, including the Victorian register, showed an increase in the rates of CP over the 1970s and 1980s, consistently seen in extremely preterm (<28wks) survivors but also in those born at term (≥37wks). Since the early 1990s, CP rates either stabilized or decreased, particularly for children born extremely preterm.
Interpretation Increases in the rates of CP during the 1970s and 1980s are in part because of the increasing survival of extremely preterm infants that occurred without a concomitant improvement in neurological outcomes. Evidence from population samples now suggests that this trend has been reversed since the mid- to late 1990s.
•Rates of CP increased during the 1970s and 1980s, partly because of the increased survival of infants born extremely preterm.
•As survival rates stabilize in extremely preterm infants, this study suggests there has been improvement in neurological outcomes over the later 1990s and early 2000s.
•Similar changes have been seen for infants born at term.
Much information can be gained from monitoring trends in the rates of cerebral palsy (CP), not only for service provision, but also for what can be learnt about aetiology. Through monitoring trends we may uncover previously unknown associations or risk factors, and we can monitor changes over time in the effects of known antecedents or interventions.
We have already learnt important lessons about the aetiology of CP through population surveillance. Although it is now more widely accepted that most cases of CP are the result of many interacting factors, previously the belief was that most CP was caused by birth trauma or intrapartum asphyxia. Efforts to reduce the incidence of CP were largely aimed at decreasing complications during labour and delivery. High hopes were held that the advent of continuous fetal heart monitoring during the 1960s and 1970s would reduce the incidence of hypoxic or ischaemic brain injury through the early detection of signs of fetal distress and intervention in the form of emergency Caesarean section. As a result of this approach, emergency Caesarean rates increased dramatically and fetal deaths during labour decreased, but the expected improvement in CP rates did not eventuate.1–3 Further research suggested that only a small proportion of CP in developed countries was the result of preventable intrapartum events.4 The aetiology of CP began to be seen more in terms of causal pathways frequently originating in the pre-conception or antenatal periods.5
Monitoring population trends in CP is not an easy task and few groups are in a position to make reliable assessments of trends over time. Most monitoring is carried out by geographically defined CP registries and requires long-term investment of resources, consistent methodologies, and large enough numbers to minimize yearly variability so that true change can be readily differentiated from random variation. Where trends in the rates of CP have been reported, the results have often been inconsistent. An early review concluded that no consistent directionality in temporal trends of CP could be extracted from studies when combined, and that the evidence was not strong enough to associate changes in the rates of CP with changes in obstetric or neonatal care.6 A second review performed in the late 1990s showed more consistency within each of four geographically defined areas but no obvious overall pattern.5 A more recent review plotted prevalence estimates of CP weighted on study size and fitted a spline smoothing line to visualize the secular trend more easily.7 The resultant plot showed a slight increase in CP prevalence over the 1980s and a decline during the 1990s. This method of consolidating data, however, is not as informative as comparing temporal trends within individual registries that have used consistent methodologies over the surveillance period.
One important causal pathway is the one involving preterm birth. An overall increase in the frequency of CP over the 1970s and 1980s reported by some CP registries was attributed to improved survival rates for extremely preterm infants who were at particular risk of germinal matrix haemorrhage and periventricular white-matter injury.8,9 Since the introduction of neonatal intensive care to Victoria, Australia, in 1968, enormous changes in the neonatal management of sick infants had their greatest impact on those born extremely preterm. The 1970s and 1980s saw the introduction of assisted ventilation, antenatal corticosteroids, and an increased willingness to offer intensive care. Exogenous surfactant was introduced in 1991, and together with greater use of antenatal steroids and more frequent in utero transfer to tertiary referral centres,10 these management strategies increased survival rates over the 1980s and 1990s. There was, however, no clear indication of improvements in neurological outcomes. This was despite good evidence from randomized controlled trials and systematic reviews for a positive effect on the rates of CP of surfactant therapy,11 antenatal dexamethasone,12 and antenatal magnesium sulphate.13 On the negative side, a systematic review of early postnatal corticosteroids in preterm infants reported an associated increased risk of an abnormal neurological examination and CP.14 By the 2000s, survival in Victoria had reached a plateau. Almost universal use of caffeine for the management of apnoea of prematurity,15 and decreasing use of postnatal steroids, were anticipated to reduce the incidence and severity of neurological injury.16
Kernicterus, brain malformations, TORCH (toxoplasmosis, other infections, rubella, cytomegalovirus, Herpes simplex virus) infections, perinatal arterial ischaemic strokes, and perinatal asphyxia are along causal pathways to CP that are more often associated with term, or near-term, birth. Overall, there have been few changes since 1983 that would appear likely to have had any major impact on the rates of CP in children born at term. In Victoria, the rate of perinatal deaths in term infants decreased, and multiple births increased, and both these could potentially have resulted in increased rates of CP. Measures to prevent kernicterus, including effective screening for rhesus incompatibility, anti-D vaccination, and phototherapy for hyperbilirubinaemia,17 as well as immunization to prevent rubella infection,18 would only have had the potential to prevent a few cases, because these causes of CP are relatively rare. Routine ultrasound has increased our capacity to detect fetal brain malformations at a time when offering termination of pregnancy is still an option, but there is no evidence to suggest a difference in long-term outcomes.19 Inherited or acquired thrombophilia in the mother or infant, placental thrombosis, infection, surgery, and the use of intravascular catheters are all factors that may contribute to the vulnerability of the fetus or infant to stroke,20 but there is little in the existing literature about the effect of these factors, or of treatments such as anticoagulant therapy, on long-term outcomes for the infant.21 Perinatal asphyxia is along an important causal pathway to CP. Electronic fetal heart monitoring and Caesarean section, in the context of obstetric catastrophes, have virtually eliminated unexpected intrapartum fetal deaths. However, no data have demonstrated that intervention based on fetal heart rate patterns or operative delivery reduces the risk of CP.22 In fact, the likelihood of CP may have increased, because already-injured infants may simply have died in utero. Neonatal extracorporeal membrane oxygenation (ECMO),23 pharmacological agents such as allopurinol24 or erythropoetin,25 hypothermia within 6 hours of birth,26 and stem-cell therapy are management strategies for preventing or ameliorating the secondary energy failure and cell death. Apart from ECMO, which has been used for two decades, and hypothermia which was introduced in Victoria in 2002, most of these therapeutic options are still in their trial phases; however, initial reports from trials of stem-cell therapy have been optimistic (Watson L, personal communication 2011).
In view of the enormous changes in societal trends and obstetric and neonatal care that have taken place in recent decades, and the important relationship that exists between causal pathways and gestational age, the aims of this study were to assess overall and gestational age-specific trends in the rate of congenitally acquired CP in Victoria, Australia, and to compare these findings with those from other populations.
The study was conducted at The Royal Children’s Hospital in Melbourne, Australia and approval was granted by the hospital’s Human Research Ethics Committee.
Data from Victoria
For this study, data were extracted from the Victorian Cerebral Palsy Register (VCPR) for 3491 individuals born with CP in Victoria between 1 January 1970 and 31 December 2004 (mean 100 registrations per birth year, range 65–131; 1963 males, 1528 females). Postneonatally acquired cases were excluded. A data set was compiled consisting of the number of individuals born with CP each year and for four groups based on gestational age at birth: extremely preterm (20–27wks), very preterm (28–31wks), moderately preterm (32–36wks), and term (≥37wks). Population data were obtained from the Victorian Perinatal Data Collection Unit, which compiles data from midwives’ statutory birth notifications. There were 2 209 758 live births (mean 63 050 per year, range 57 769–75 498) for the 35 years of the study. Although data about live births were available from 1970, gestational age-specific data were only available from 1983. Only 15 of 2400 individuals with CP born since 1983 had missing data on gestational age.
Review of the literature
Searches were conducted on MEDLINE for publications with data on trends in the rates of CP over time. The search criteria used were ‘cerebral palsy’ as a main subject heading, ‘epidemiology’ as a subheading, and ‘trends’ as a keyword. Bibliographies, review articles, and registry reports were also used to source new data. For inclusion in the review, data were required to be from a geographically defined CP registry in a developed country that had used consistent methods to obtain at least 15 years of consecutive data since 1970. A consistent type of denominator of at least 20 000 was necessary for the calculation of rates. In addition, live births as the denominator was a requirement for trends in CP following preterm birth, as the use of neonatal survivors would have needed to be interpreted differently.5 Cerebral palsy rates were extracted from more than one publication per registry when published as separate instalments over time, and, when birth years were grouped, the middle year was used as the reference year.
Analyses were conducted using Stata 11 software (StataCorp 2009, College Station, TX, USA). Rates were calculated per 1000 live births for all cases of CP and for each gestational age group using gestational age-specific population data as the denominator; these were tabulated and plotted by year of birth. Fractional polynomial curves were fitted to enable better visualization of trends over time.
Reports of trends in the rates of CP were found from nine CP registries that fulfilled the study’s inclusion criteria, including the VCPR.1,5,8,27–39 Details are shown in Table I.
Table I. Registries of cerebral palsy (CP) included in the systematic review of trends in rates for all CP or gestational age-specific CP rates
Over the period 1970 to 2004, there was a gradual increase in the rate of CP in Victoria, with most of the increase occurring during the 1970s and 1980s (Fig. 1). In addition to the VCPR, reports of trends in the overall rates of CP were found from five CP registries. All regions showed increases in their rates of CP over the 1970s and early 1980s, and, since the early 1990s, the data suggest that rates have either stabilized, as in Victoria, or have been decreasing (Fig. 1). Although there were noticeable differences in prevalence, very similar trends were seen throughout the 1970s, 1980s, and 1990s for Western Australia and Victoria.
Gestational age-specific trends
Between 1983 and 2004, data from Victoria showed a gradual increase in the rate of CP in infants born at term, reaching a peak in the mid-1990s. Trends in CP rates for infants born at term were also included from three other registries: eastern Denmark, western Sweden, and Western Australia. A similar pattern was seen in the Western Australian data, and an earlier peak occurred in western Sweden (Fig. 2).
Moderately preterm birth
A very small decrease in the rates of CP for moderately preterm infants (32–36wks) was seen for the first half of the study period in Victoria. Data from Western Australia and western Sweden also showed similar small reductions in the rates of CP over the same period, whereas data from eastern Denmark showed dramatic reductions for this group of infants (Fig. S1, supporting information published online).
Very preterm birth
No directional trend in the rates of CP following very preterm birth (28–31wks) was seen in the Victoria data for the birth years 1983 to 2004. In contrast, data from Western Australia showed decreasing rates, at least for part of the study period (Fig. S2, supporting information published online).
Extremely preterm birth
An increase in the rate of CP per 1000 live births was seen over the first half of the study period in Victoria for children born extremely preterm (<28wks). The birth cohort prevalence of CP peaked in the mid-1990s and has since been decreasing. A similar peak was seen in data from Alberta, Canada, western Sweden, and Western Australia (Fig. 3).
The VCPR collects information on children with CP from a large population base of around 65 000 births per year. This gives us the advantage of having higher numbers of cases for subgroup analyses of trends. We also have good gestation-specific population data from 1983 onwards. Our overall CP rate of 1.8 per 1000 live births is within the expected range of 1.5 to 2.5 per 1000 live births,7 and the methods used to collect data have been consistent over time.
The results of this study indicate that overall CP rates in Victoria rose over the 1970s, 1980s, and 1990s from around 1.0 per 1000 live births in 1970 to around 1.8 at the beginning of the current decade. This difference in birth cohort prevalence corresponds to around 50 extra registrations each year in Victoria. Although it is difficult to predict trends into the future, there is a suggestion that the rates may have stabilized. To varying degrees, the same general increase in rates was seen in data from the other reviewed CP registries for the 1970s and 1980s and there is a suggestion that the rates may be decreasing, although more recent data are less consistent.
It can be difficult to distinguish real changes in rates from variations in levels of ascertainment. In Victoria, data on individuals born in the 1970s were collected retrospectively and ascertainment during this time may have been less complete than in later decades. Even in the 1990s and 2000s, the numbers of extremely preterm infants diagnosed with CP and registered with the VCPR were lower than the numbers obtained from the Victorian Infant Collaborative Study where infants were followed up and assessed at age 2.16 It is possible that not all children with CP were identified by the VCPR, but it is also possible that different criteria were used to apply the label of CP, or that abnormal neurological signs resolved between 2 years and 5 years of age. Evidence from California suggests that the latter situation is not uncommon. The Californian study demonstrated that up to 50% of children diagnosed with CP at 1 year of age were free of impairment at age 7.40 It is unknown whether inconsistent ascertainment has affected trends in prevalence estimates from any of the registries included in this study.
In regions where CP rates increased over the 1980s and 1990s, the increase was often attributed to improved survival of extremely preterm infants, who are at greater risk of developing CP. Data from the VCPR support this notion. There was a rise in the rates of CP in extremely preterm infants throughout the 1980s and the first half of the 1990s, but since then rates have decreased. Neonatal follow-up programs in Victoria have published data that support the lack of improvement in CP rates over the 1980s and 1990s.16 Furthermore, data from Sweden and Canada showed a peak similar to that seen in Victoria; however, rates reported by the Western Australian Register have continued to rise into the 2000s. Relatively wide variations have been reported between longitudinal cohorts in the prevalence of CP following extremely preterm birth, and this variation has mainly been attributed to changing mortality rates and to different clinical practices such as aggressiveness of intervention and variations in the use of new therapies.41,42
Trends in the rates of CP among preterm infants may not account for all of the increase in rates of CP seen over the period of this study. In Victoria, there was also a gradual increase in the rates of CP in infants born at term, peaking in the mid-1990s and mirroring trends in the rates for all CP. A change of 0.1 per 1000 live births in term-born infants equates to around five extra cases of CP in Victoria per year. The fact that the same pattern was seen for all CP, and that the pattern was mirrored in the data from Western Australia, supports the concept of a small but real change. A similar trend was seen in data from western Sweden, although the peak occurred earlier. It is unknown whether the observed increase was a general one, or whether it involved particular causal pathways.
Little attention has been paid to rates of CP in infants born between 28 weeks and 36 weeks gestational age, and less is known about whether changes in neonatal practice have resulted in improved outcomes for these later-born preterm infants. In this study, relatively little change in rates for these infants was observed in Victoria, but this relative stability was not universally seen in data from the other regions included in this review. In contrast to Victoria, the rates of CP in very preterm and moderately preterm infants from Western Australia and western Sweden were substantially higher than rates from Victoria, and tended to decrease over the 1980s and 1990s to nearer the Victorian rates.
The rate of non-postneonatally acquired CP in Victoria increased during the 1980s and early 1990s. This was partly due to the increasing survival of extremely preterm infants, which apparently occurred without a concomitant improvement in neurological outcomes, and because of a small rate increase in infants born at term. Importantly, there is now evidence to suggest that this trend has been reversed since the mid- to late 1990s both for infants born at term and those born extremely preterm. These findings are supported by data from some population samples in other developed countries.
We are grateful to Dr Rod Hunt who reviewed the manuscript, Anna Lanigan and Tess Lionti who collected data for the VCPR, and the Victorian Perinatal Data Collection Unit that provided population birth data for this study. We also acknowledge the support of the Victorian Medical Insurance Agency Ltd (Professional Services Australia), and the Victorian Department of Health who provided funding for the Victorian Cerebral Palsy Register. The first author’s doctoral research is supported by a scholarship (2009–2011) from the Australian National Health and Medical Research Council.