Postnatal corticosteroids and neurodevelopmental outcomes in extremely low birthweight or extremely preterm infants: 15-year experience in Victoria, Australia
- Jeanie Ling Cheong1,2,3,
- Peter Anderson2,4,
- Gehan Roberts2,5,
- Julianne Duff6,
- Lex W Doyle2,3,
- Victorian Infant Collaborative Study Group
- 1Neonatal Services, Royal Women's Hospital, Parkville, Victoria, Australia
- 2Victorian Infant Brain Studies, Murdoch Children's Research Institute, Parkville, Victoria, Australia
- 3Obstetrics and Gynaecology Department, University of Melbourne, Parkville, Victoria, Australia
- 4Psychology Department, University of Melbourne, Parkville, Victoria, Australia
- 5Paediatrics Department, Royal Children's Hospital, Parkville, Victoria, Australia
- 6Royal Women's Hospital, Parkville, Victoria, Australia
- Correspondence to Jeanie Ling Cheong, Neonatal Services, Royal Women's Hospital, Level 7, Newborn Research, 20 Flemington Road, Parkville, Victoria 3052, Australia;
Contributors Jeanie Cheong – conceptualisation of study, data interpretation, drafting and revising the manuscript. Peter Anderson – data interpretation, drafting and revising the manuscript. Gehan Roberts – data interpretation, drafting and revising the manuscript. Julianne Duff – data interpretation, drafting and revising the manuscript. Lex Doyle – conceptualisation of study, data interpretation, drafting and revising the manuscript.
- Received 8 November 2011
- Accepted 20 February 2012
- Published Online First 9 June 2012
Objective Postnatal corticosteroids (PCS) are used to prevent or treat bronchopulmonary dysplasia (BPD) in extremely low birthweight (ELBW; <1000 g) or extremely preterm (EPT; <28 weeks) infants. In the early 2000s, concerns were raised about increased risks of cerebral palsy (CP) in association with PCS, which may have affected prescribing of PCS, and influenced rates of BPD, mortality or long-term neurosensory morbidity. Our aim was to determine the changes over time in the rates of PCS use and 2-year outcomes in ELBW/EPT infants in Victoria, Australia.
Design All ELBW or EPT infants born in Victoria, Australia in three distinct eras (1991–92, 1997 and 2005) who were alive at 7 days were included. Rates of PCS use, rates of BPD (oxygen dependency at 36 weeks' corrected age), death before 2 years of age, CP and major disability (any of moderate/severe CP, developmental quotient <−2 SD, blindness or deafness) were contrasted between cohorts.
Results The rate of PCS use and the dose prescribed diminished significantly in 2005 compared with earlier eras, but the rate of BPD rose. Non-significant changes in the rates of mortality over time were mirrored by non-significant changes in the rates of CP or major disability. Combined outcomes of mortality with either major disability or CP were similar over the three eras.
Conclusions PCS use decreased in 2005 compared with earlier eras, and was accompanied by a rise in BPD, with no significant changes in mortality or neurological morbidity.
What is already known on this topic
The link between postnatal corticosteroid use and cerebral palsy in preterm infants has altered prescribing practices.
It is unclear whether subsequent rates of bronchopulmonary dysplasia or adverse neurodevelopmental outcomes have been affected due to conflicting data.
What this study adds
Despite decreased postnatal corticosteroid use as well as total dose exposure, the rates of mortality or adverse neurodevelopmental outcomes in a regional cohort remain unchanged.
Bronchopulmonary dysplasia rates over that time were increased.
Factors other than postnatal corticosteroids are likely to contribute to the persistent adverse outcomes in extremely preterm or extremely low birthweight infants.
Postnatal corticosteroids (PCS) have been widely used in preterm infants to prevent or treat bronchopulmonary dysplasia (BPD),1 ,2 also known as chronic lung disease. The commonest PCS in this circumstance is dexamethasone, although hydrocortisone and other corticosteroids have also been used. In 2002, the American Academy of Pediatrics (AAP) published a consensus statement cautioning the use of PCS to either treat or prevent BPD as death rates were unchanged, and any benefit of reduced BPD was potentially offset by an increased risk of cerebral palsy (CP).3 The rates of PCS use worldwide subsequently decreased.4,–,8 The trend to decrease use of PCS resulted in increased rates of BPD in some studies,5 ,6 ,9 but similar rates in others.4 ,7 ,10
Recent reviews have concluded that PCS given early to prevent BPD (<8 days after birth) demonstrated benefits with early extubation and decreased BPD; however, the rate of adverse long-term neurodevelopmental outcomes increased, particularly CP.1 ,11 ,12 Late PCS (>7 days after birth) may reduce mortality but not significantly increase the risk of adverse neurodevelopmental outcomes.2 The effect of PCS on the combined outcome of death or CP varies with BPD risk, suggesting that babies at very high risk of BPD might benefit in the long term,13 ,14 but there are no conclusive trials that have been able to demonstrate such improvement.
Short of data from randomised-controlled trials, data from cohort studies might provide insight into patterns of prescribing of PCS and long-term outcomes. In the state of Victoria, Australia, there have been three recent geographically determined cohort studies which have spanned the time from early and accelerating rates of PCS prescribing in the 1990s, to the lower rates in the 2000s, after the AAP statement was released.
The aims of this study were to determine changes over three distinct eras, 1991–1992, 1997 and 2005 in PCS treatment, and of BPD, survival and neurodevelopmental outcomes at 2 years of age in extremely low birthweight (ELBW) or extremely preterm (EPT) infants in Victoria, Australia. It was hypothesised that over the three eras, (1) the rate of PCS use would fall, (2) the rates of BPD and mortality would rise and (3) the rate of neurological disability in survivors would fall, thus resulting in a steady rate of survival free of neurological disability.
This retrospective analysis included all ELBW (birthweight<1000 g) or EPT (gestational age <28 weeks) infants live born in the state of Victoria, Australia, from three distinct eras, that is, those born during the calendar years of 1991–1992, 1997 and 2005, who were alive at 7 days.15 ,16 Details of recruitment have been described previously.17
Perinatal data collection
Perinatal and neonatal data were collected by research nurses by chart review. Over the study period, PCS to treat or prevent BPD were prescribed almost exclusively after the first week of life. The occasional use of PCS, usually hydrocortisone, in the first week to support hypotension was not considered as ‘having received PCS’. Consequently, data on all outcomes were included only for those infants who survived the first week after birth. PCS use after the first week was documented in all eras; data on dose, age of commencement and total duration were collected for the first and last eras only. The PCS prescribed was mostly dexamethasone, either parenteral or oral. For the few occasions when other corticosteroids were used, such as hydrocortisone or prednisolone, the dose was converted to dexamethasone-equivalent amounts, assuming relative potencies of dexamethasone 25, prednisolone 4 and hydrocortisone 1. Inhaled corticosteroids were rarely used and not included in the analysis. BPD was defined as oxygen dependency at 36 weeks. Data on the duration of oxygen therapy, hospitalisation before discharge home or death and the rate of home oxygen therapy were recorded.
Two-year outcome measures
Survival was determined at 2 years of age, corrected for prematurity, when survivors were assessed by paediatricians and psychologists blinded to all perinatal details, including gestational age at birth, as the children were assessed concurrently with normal birthweight term controls.15 ,16 The following impairments were evaluated: CP, blindness (visual acuity of <6/60 in the better eye), deafness (hearing loss requiring amplification or worse) and developmental delay. Criteria for diagnosis of CP included abnormal muscle tone and delays in motor control and function.18 ,19 Development was assessed using the Bayley Scales of Infant Development, using the first published version for the 1991–1992 cohort,20 the revised version for the 1997 cohort,21 and the third edition for the 2005 cohort.22 Developmental delay was classified according to the developmental quotient computed in SD scores relative to the mean for contemporaneous normal birthweight term controls for each of the respective ELBW/EPT groups on (1) the Mental Developmental Index of the original and revised version Bayley Scales of Infant Development,20 ,21 and (2) according to the Cognitive or Language Composite scores on the Bayley-III.22 Major neurological disability comprised any of a DQ<−2SD, non-ambulant CP (equivalent to a Gross Motor Classification System19 level II or worse), blindness or deafness.
The Research and Ethics Committees at the Royal Women's Hospital, Mercy Hospital for Women and Monash Medical Centre, Melbourne approved the follow-up studies.
Data were analysed using SPSS for Windows version 17.0 (SPSS Inc., Chicago, Illinois, USA). Differences in proportions between groups were compared using χ2 analysis, and OR and 95% CI calculated. Differences in means between groups were compared with t-tests, or Mann–Whitney U tests if data were skewed. Where differences were compared across three eras simultaneously, we chose to contrast 1991–1992 with 1997, and 2005 with 1997, as 1997 was the pivotal era before the change in clinical practice; for continuous variables, we used linear regression and calculated mean differences and 95% CIs between eras, and for dichotomous outcomes, we used logistic regression and calculated ORs and 95% CIs for the various outcomes. The preceding analyses were both unadjusted and then adjusted for gestational age at birth, birth weight and gender.
The numbers of live births free of lethal anomalies in each era were 552 in 1991–1992, 296 in 1997 and 342 in 2005, and the proportions who were alive at the end of the first week were 63%, 75% and 75%, respectively. The gestational age, birth weight and gender distribution of infants alive at the end of the first week across eras were similar, with the exception that those in 1991–1992 were heavier than in 1997 (table 1).
The proportion of infants treated with PCS decreased significantly in 2005 compared with 1997 (table 1). Overall, compared with those not treated with PCS (n=562), those treated with PCS (n=261) were more immature (completed weeks, mean(SD); PCS 25.5(1.5), no PCS 27.0(1.9); mean difference −1.5, 95% CI −1.8 to −1.3), lighter at birth (g, mean(SD); PCS 766(164), no PCS 893(166); mean difference −127, 95% CI −151 to −103), and a larger proportion were male (PCS 59.4%, no PCS 46.1%; OR 1.71, 95% CI 1.27 to 2.30), with little evidence of a difference between eras in these characteristics (data not shown).
Comparing treatment regimens in 1991–1992 with 2005, those in 2005 were treated with shorter courses and much lower PCS doses than in 1991–1992, but the postnatal age and the postmenstrual age when treatment was started were similar (table 2).
Female gender, increasing birth weight and increasing gestational age were all associated with shorter durations of oxygen therapy and hospitalisation, and lower rates of being discharged home on oxygen (table 3). After adjustments for these variables, the durations of oxygen therapy and of the primary hospitalisation were longer in 2005 compared with 1997, although the proportions discharged home on oxygen were similar (table 3).
Follow-up rates of survivors at 2 years were 97.7% (1991–1992), 99.0% (1997) and 95.5% (2005).15 ,16 There were small fluctuations in the rates of CP or major disability in opposite directions to the small changes in mortality (table 1). However, the rates of major disability diminished in 2005 compared with 1997. The rates of CP, and the combined outcomes of death or CP, or death or major disability were similar between eras.
Increasing maturity and birth weight were independently associated with lower rates of PCS use, BPD and mortality, and in the combined outcomes of death with CP or major disability, but not with CP or disability alone (table 4). Male gender was independently associated with higher risks of PCS use, BPD, mortality and major disability, and in the combined outcomes of death with CP or major disability, but not with CP alone (table 4). After adjustment for confounding variables, the decreased rate of PCS use and the increase in BPD in 2005 compared with 1997 remained statistically significant, but the decrease in major disability in 2005 did not. However, the decrease in mortality in 1997 compared with 1991–1992 became statistically significant. All other conclusions were unchanged.
The rate of PCS treatment to prevent or treat BPD in ELBW/EPT infants in Victoria, Australia, decreased substantially in 2005 compared with 1997, mirrored by an increase in BPD rates, as well as longer durations of oxygen therapy and hospitalisation. The dose and duration of PCS were both considerably reduced in 2005 compared with 1991–1992. As expected, increasing maturity and birth weight, and female gender were associated with lower rates of PCS treatment, as well as morbidity (including shorter durations of oxygen therapy, hospitalisation and lower rates of home oxygen therapy) and mortality. Rates of mortality, CP and major disability, and for combined outcomes of death with CP or major disability were similar over time, after adjustment for confounding variables. The observation that neurological morbidity or mortality in ELBW/EP infants remained unchanged despite the decrease in PCS therapy over this 15-year period suggests that other factors may be contributing to the neurological morbidity faced by survivors of neonatal intensive care.
Several publications, mainly from neonatal networks in North America and one from Israel, have reported changes over time in PCS use and morbidity including BPD and neurodevelopmental impairment in ELBW or EPT infants.4,–,9 Most studies have reported a decrease in PCS use in the mid-2000s in comparison with the mid-to-late 1990s.4,–,8 The rates ranged from ≈25% to 65% in the late 1990s, to 8% to 23% in the mid-2000s. Our PCS use fell from 37% in 1997 to 23% in 2005, which is within these reported ranges. One exception to this trend was a single centre report where the rate of PCS use increased from 9% in 1985–1986 to 39% in 2005–2006.9 Compared with more immature cohorts in other studies, this study had an inception cohort of infants ≤30 weeks and the comparison eras were different from most other studies. The trends in PCS use may simply reflect practices unique to this single centre, where rates of PCS use in the mid-2000s were higher compared with other studies.
Our finding of increased BPD rates with a drop in PCS use was consistent with some reports. Shinwell et al reported that the adjusted OR for BPD was 1.4 times in 2003–2004 compared with 1997–1998, in conjunction with a fall in PCS use from 23.5% to 11%.5 Our rates of BPD were in the vicinity of 45%, which is higher than that of another report by Yoder et al, but this may have been explained by the more immature nature of our inception cohort in comparison with that study where infants ranged between 23 and 32 weeks' gestation at birth.6 Nonetheless, there are other reports of unchanged BPD rates over a similar time period.4 ,7 ,8 In a single centre study, Wilson-Costello reported a reduction of PCS use in ELBW infants from 53% in 1990–1999 to 23% in 2000–2002.8 The rates of BPD were steady at 46% and 51%, respectively, which is similar to our increased rate in 2005. Another study reported BPD rates of 68.4% and 71.4% between 1999–2001 and 2002–2004, but the cohort comprised the highest risk infants born at <25 weeks.4 A large observational study of three large neonatal networks in USA and Canada reported rates of BPD that were unchanged at around 25% between 2001 and 2003 despite a reduction in PCS from 25% to 8% in very low birthweight infants (<1500 g).7 All reports used a similar definition of BPD to our study, that is, oxygen dependency at 36 weeks postmenstrual age.
Neurodevelopmental morbidity and mortality in many cohorts have largely remained unaltered during the period of interest in association with reductions in PCS use, as confirmed by our regional study. One single centre study reported an improvement in rates of neurodevelopmental impairment in their ELBW cohort from 35% (1990–1999) to 23% (2000–2002).8 In contrast, using similar definitions, our rates of major neurodevelopmental disability remained around 15%–25% throughout the study duration.
There is limited information from observational studies about the changes in PCS total dosages over time. PCS dosing regimens that are commonly in use range from a total dexamethasone dose of 0.8923 to 8 mg/kg.24 In Victoria, Australia, there has been a move toward using a low-dose dexamethasone regimen which was reflected in the finding of a total dose reduction from a mean of 7.7 mg/kg in 1991–1992 to 2.0 mg/kg dexamethasone equivalent in 2005.
Our study is one of the few studies to report regional data with high follow-up rates and similar standardised outcome assessments blinded to perinatal information over a 15-year period. However, as with any observational study, we are only able to report temporal trends and associations, but not causality. Regardless, our findings of neurodevelopmental outcome rates that are unchanged with a reduction in both PCS use and total dose exposure are consistent with the Cochrane Review of late corticosteroid administration, and studies from other neonatal networks.2 ,14 We can only speculate that the increased consumption of resources in 2005 compared with 1997 may be attributable to the lower rate of PCS in that era.
In conclusion, despite concerns about PCS in the causal pathway of adverse long-term neurodevelopmental outcomes in ELBW/EPT infants, there is increasing evidence from observational studies that a reduction in PCS use has not been associated with dramatic improvement in neurodevelopmental outcomes. PCS continues to be used in around one-in-four ELBW/EPT survivors after the first week of life in Victoria, Australia, but at a much reduced total dose in 2005 compared with the early 1990s. It would be reasonable to extrapolate that PCS have a role in the management of those at high risk of BPD, without significantly affecting their outcome.13 Future directions of research would include definitive randomised-controlled trials of dexamethasone compared with placebo for the treatment of BPD targeted at infants at high risk of BPD. However, given the lack of equipoise, the feasibility of such trials would be difficult.23 There is also a role for randomised-controlled trials of alternative PCS such as hydrocortisone in comparison with dexamethasone in the efficacy of treatment or prevention of BPD, as well as adverse neurodevelopmental effects.
Collaborators Participants: Convenor; Lex W Doyle MD, FRACP. Collaborators (in alphabetical order); Peter J Anderson PhD Catherine Callanan RN, Elizabeth Carse FRACP, Margaret P Charlton M Ed Psych, Mary-Ann Davey PhD, Noni Davis FRACP, Julianne Duff FRACP, Rod Hunt PhD, FRACP, Cinzia de Luca PhD, Marie Hayes RN, Esther Hutchinson DPsychClinNeuro, Elaine Kelly MA, Marion McDonald RN, Gillian Opie FRACP, Gehan Roberts PhD, FRACP, Michael Stewart FRACP, Andrew Watkins FRACP, Amanda Williamson MA, Heather Woods RN.
Funding NHMRC project grants.
Competing interests None.
Ethics approval Royal Women's Hospital, Mercy Hospital for Women, Monash Medical Centre.
Provenance and peer review Not commissioned; externally peer reviewed.