Authors: Cassidy Du Berry, Christopher Nesci, Jeanie L.Y. Cheong, Tara FitzGerald, Rheanna Mainzer, Sarath Ranganathan, Lex W. Doyle, Elianne J.L.E. Vrijlandt, Liam Welsh
Categories: Articles, Expiratory airflow, Moderate-late preterm, Preterm, Pulmonary function, Spirometry
Source: eClinicalMedicine
Moderate-late preterm (MLP; 32 to <37 weeks’ gestation) birth is associated with reduced expiratory airflow during child, adolescent and adult years. However, some studies have reported only minimal airflow limitation and hence it is unclear if clinical assessment in later life is warranted. Our aim was to compare maximal expiratory airflow in children and adults born MLP with term-born controls, and with expected norms.
We systematically reviewed studies reporting z-scores for spirometric indices (forced expired volume in 1 second [FEV1], forced vital capacity [FVC], FEV1/FVC ratio and forced expiratory flow at 25-75% of FVC [FEF25-75%]) from participants born MLP aged five years or older, with or without a term-born control group from 4 databases (MEDLINE, CINAHL, Embase, Emcare). Publications were searched for between the 22^nd^ of September 2021 to the 29^th^ of September 2021. A meta-analysis of eligible studies was conducted using a random effects model. The study protocol was published in PROSPERO (CRD #42021281518).
We screened 4970 articles and identified 18 relevant studies, 15 of which were eligible for meta-analysis (8 with term-born controls and 7 without). Compared with controls, MLP participants had lower z-scores (mean difference [95% confidence interval] I^2^) for FEV1: -0.22 [-0.35, -0.09] 49.3%, FVC: -0.23 [-0.4, -0.06] 71.8%, FEV1/FVC: -0.11 [-0.20 to -0.03] 9.3% and FEF25-75%: -0.27 [-0.41 to -0.12] 21.9%. Participants born MLP also had lower z-scores, on average, when compared with a z-score of 0 (mean [95% CI] I^2^) for FEV1: -0.26 [-0.40 to -0.11] 85.2%, FVC: -0.18 [-0.34 to -0.02] 88.3%, FEV1/FVC: -0.24 [-0.43 to -0.05] 90.5% and FEF25-75%: -0.33 [-0.54 to -0.20] 94.7%.
Those born MLP had worse expiratory airflows than those born at term, and compared with norms, although reductions were modest. Clinicians should be aware that children and adults born MLP may be at higher risk of obstructive lung disease compared with term-born peers.
This work is supported by grants from the National Health and Medical Research Council (Centre of Research Excellence #1153176, Project grant #1161304); Medical Research Future Fund (Career Development Fellowship to J.L.Y Cheong #1141354) and from the Victorian Government's Operational Infrastructure Support Programme. C. Du Berry's PhD candidature is supported by the Melbourne Research Scholarship and the Centre of Research Excellence in Newborn Medicine.
Keywords: Preterm, Moderate-late preterm, Pulmonary function, Spirometry, Expiratory airflow
Preterm birth is associated with increased respiratory morbidity during infancy, and impaired expiratory airflow later in life.1, 2, 3 The incidence of moderate late-preterm (MLP) birth, defined as birth between a gestation of 32 to < 37 weeks, has steadily increased over the previous decade. For example, the Australian MLP preterm birth rate has increased by a relative 6.2% between 2009 to 2019, accounting for 6.9% of all live births in 2019.^4^^,^^5^ A similar pattern has been observed in the United States (US), where the MLP preterm birth rate has risen by 3.7% over the same period.^6^ The increase in the MLP birth rate has been simultaneous with advancing maternal ages and better obstetric surveillance.4, 5, 6 As infants born MLP vastly outnumber very preterm (VP; 28 to <32 weeks’ gestation) and extremely preterm (EP; <28 weeks’ gestation) survivors, morbidity in those born MLP may result in substantial economic and healthcare burden.^7^
Compared with infants born at term, infants born MLP are at an increased risk of developing adverse respiratory sequelae, such as asthma and recurrent wheeze.^1^^,^^8^ It is becoming increasingly evident that birth during the 32 to <37 week gestational window interrupts a time critical period of rapid in utero respiratory growth. During the saccular and alveolar phase of gestation, acinar structures undergo a period of maturation which is characterised by peripheral airway enlargement, decreasing air-space wall thickness, and increasing alveolar surface area.^9^^,^^10^ Deviation from this finely programmed series of normal lung development may cause alterations in pulmonary mechanics during infancy, resulting in an overly compliant chest wall, reduced expiratory airflow and increased airway resistance at birth.11, 12, 13
Structurally immature bronchial and parenchymal networks, in addition to incurred respiratory morbidity during infancy, may have adverse effects on pulmonary function later in life.^3^ Recent data from prospective birth cohort studies have reported an association between MLP birth and reduced expiratory airflow during school age and adolescent years, which persists into early adulthood.14, 15, 16 However, studies have reported varying degrees of airflow limitation and therefore the trajectory of expiratory airflow for infants born MLP remains unclear. Our study aims to compare maximal expiratory airflows in children and adults born MLP with term-born controls, and with expected norms.
Our study was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines^17^ and the study protocol is published in PROSPERO International Prospective Register of Systematic Review (CRD #42021281518).
Published studies met inclusion criteria if they reported z-scores for the spirometric indices of forced expiratory volume in 1 second (FEV1), forced vital capacity (FVC), FEV1/FVC or forced expiratory flow at 25-75% of FVC (FEF25-75%) from participants aged 5 years or older who were born MLP. If studies included data obtained from age-matched term-born controls, comparison was made between MLP and term-born participants. A language restriction was not placed on the literature search, although to sufficiently assess the quality of each study we required the full text to be available in English. Studies published between January 1984 and September 2021 were eligible for inclusion in the review. Studies that reported outcomes in absolute units or percent predicted, or reported pulmonary function measurements other than spirometry were ineligible for meta-analysis but were eligible for review. Studies were excluded if data obtained from MLP participants could not be differentiated from EP or VP participants.
Relevant studies were identified by searching 4 electronic databases (MEDLINE, CINAHL, Embase and Emcare) between the 7^th^ of September 2021 to the 21^st^ of September 2021. A repeat search was conducted on 28 January 2022. We conducted 3 discrete searches for each database using terms specific to preterm birth and pulmonary function testing. The complete search strategy is available in e-Appendix #1 of the online supplementary material. Using the inclusion criteria, two independent authors (C.D.B and C.N) removed duplicates, screened the titles and abstracts of retrieved articles and obtained full-text articles. A third author (L.W) resolved any dispute. When studies reported outcomes from the same cohort in two separate articles, data from the first time point were used. Explanations for study exclusion at full-text screening are available in e-Appendix #2.
Z-scores of spirometric indices from eligible studies were extracted for meta-analysis by two authors (C.D.B and C.N). A third author (L.W) verified the data extracted. Authors were contacted for aggregate-level data if a study did not report indices of spirometry as z-scores derived from the Global Lung Initiative (GLI) reference equations.^18^ Authors were also contacted if a study reported indices of spirometry obtained from a combination of both VP and MLP born participants. In such instances, only aggregate-level data from MLP born participants were requested. Additionally, authors were contacted when a spirometric index of interest was recorded but not reported. Responses from contacted authors were considered until January 11, 2022. For longitudinal studies that presented cohort data at multiple time points, data from the first time point were used.
STATA version 17.0 (StataCorp, 2021) was used to analyse data. Meta-analysis was conducted using the metan statistical analysis package.^19^ Birth and perinatal characteristics of participants, sample size, pulmonary function tests performed and spirometry reference values were summarised for each study (Table 1). For each study, the mean z-score and standard deviation for FEV1, FVC, FEV1/FVC and FEF25-75% in addition to number of participants were extracted for MLP and term-control groups.^20^ In cases where the mean and standard deviation were not reported, the median and interquartile range (IQR) were used instead if sample sizes were larger than 25 participants in each group.^21^ If z-scores were presented for separate gestational age (GA) groups within the 32 to <37 week gestational period, means and standard deviations for each group were combined in accordance with the Cochrane handbook for systematic reviews.^22^
Separate meta-analyses were performed for studies that included term-born controls only and for all studies. For studies with term-born controls, mean differences in z-score of FEV1, FVC, FEV1/FVC and FEF25-75% between MLP participants and term-born controls were estimated. For all studies, mean z-scores from all MLP participants were estimated and compared with the population mean (i.e., a z-score of 0). An assumption was made that the population mean used for each study reflected the most up to date reference equations available at the time of publication. Overall estimates were obtained using a random effects model, with between-study variability estimated using the Empirical Bayes method. Results from the meta-analysis are presented with study weight (%), 95% confidence interval (CI), Z values, P values and heterogeneity statistics (I^2^). A funnel plot and Egger's test were used to assess the risk of publication bias when there were at least 10 or more studies included in the meta-analysis.^23^^,^^24^ If outcomes could not be obtained as GLI z-score, studies were still eligible for inclusion in the meta-analysis but were subject to a sensitivity analysis.
Studies of cohort or case control design were assessed using the Newcastle-Ottawa Quality Assessment Scale (NOS) for cohort studies or the NOS for case control studies.^25^ For cohort studies, we considered that participants lost to follow-up were unlikely to introduce bias if follow-up rates were ≥80%, or between 70% and 80% with an accompanying statement describing those lost to follow up. Cross sectional studies were assessed using the Joanna Briggs Institute (JBI) critical appraisal checklist for analytical cross sectional studies,^26^ while version 2 of the Cochrane risk-of-bias tool was utilised to evaluate studies of randomised controlled trial (RCT) design.^27^
The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. All authors confirm that they had full access to all the data in the study and accept responsibility to submit for publication.
The PRISMA 2020 flow diagram^17^ for searched, identified, screened, and included records is presented in Figure 1. 4970 articles were identified from selected electronic databases after duplicates were removed. After screening the title and abstract of each article, 56 full-text articles were screened for eligibility. Once the responses from contacted study authors were collated, 18 articles satisfied the inclusion criteria.14, 15, 16^,^28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42 15 studies reported expiratory airflow data from MLP participants,14, 15, 16^,^31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42 eight of which compared MLP participants with term-born controls.14, 15, 16^,^37, 38, 39, 40, 41, 42 Three studies were not eligible for meta-analysis, with two unable to provide z-score data,^29^^,^^30^ and one with indices of impulse oscillometry only.^28^ Of the 15 studies eligible for meta-analysis, 15 reported a z-score for FEV1, FVC, and FEV1/FVC with 13 reporting FEF25-75%. Though spirometry indices were not published as GLI z-score in five of the included studies,^14^^,^^33^^,^^35^^,^^38^^,^^39^ these data were subsequently provided by four of the respective study authors.^33^^,^^35^^,^^38^^,^^39^ We obtained aggregate-level MLP data from four studies that reported a combination of MLP and VP expiratory airflow data.^31^^,^^33^^,^^38^^,^^42^ One study presented data as median and IQR.^41^ A repeat literature search on 28 January 2022 did not reveal any additional studies eligible for inclusion in the review. Across the eight studies which reported spirometric indices from MLP and term-born participants that were eligible for meta-analysis, there were a total of 847 MLP participants and 8,209 controls.14, 15, 16^,^37, 38, 39, 40, 41 An additional 819 MLP born participants were included in the seven studies without term-born participants,31, 32, 33, 34, 35, 36^,^^42^ resulting in 1,666 MLP born participants in total.
Figure 1 PRISMA 2020 flow chart of the searched, identified and included studies. Abbreviations: MLP = moderate-late preterm; VP = very preterm; EP = extremely preterm;*Some studies satisfied more than 1 exclusion criterion.
Birth and perinatal characteristics of participants, sample size, pulmonary function tests performed and spirometry reference values used are presented in Table 1. Included studies were published between 1993 and 2021 and varied in study design; observational cohort (n = 8),14, 15, 16^,^^29^^,^^37^^,^^38^^,^^40^^,^^42^ cross-sectional (n = 8),^28^^,^^30^^,^^31^^,^^33^^,^^34^^,^^36^^,^^39^^,^^41^ case control (n = 1)^32^ and RCT (n = 1).^35^ Three studies reported longitudinal spirometry measurements.^14^^,^^15^^,^^37^ The age of participants born MLP ranged from 5 to 25 years; 15 studies reported pulmonary function outcomes obtained from school-aged children and adolescents,^14^^,^^15^^,^28, 29, 30, 31, 32, 33, 34^,^^36^^,^39, 40, 41, 42 while three reported outcomes obtained from adults.^16^^,^^37^^,^^38^ Most studies included participants that were born between 32 to <37 weeks gestation (n = 14, 78%).14, 15, 16^,^^28^^,^^30^^,^^31^^,^^33^^,^^34^^,^36, 37, 38, 39, 40^,^^42^ Three studies reported outcomes from those born 32 to <36 weeks^29^^,^^32^^,^^35^ and another between weeks 35 to <37.^41^ Studies were conducted mostly in high income countries (n = 16, 89%), and across a range of geographical locations; Europe (n = 13),^14^^,^^15^^,^^29^^,^^30^^,^^32^^,^34, 35, 36, 37^,^^39^^,^^40^^,^^42^ North America (n = 2),^31^^,^^38^ South America (n = 2)^28^^,^^33^ and the Middle East (n = 1).^41^
The highest attained rating using the NOS for cohort studies was 8/9 stars, achieved by one study.^14^ The median score was seven stars, with a range of 5 to 7. Most participants from included cohort studies were representative of children and adults born MLP in the community (n = 6, 75%),14, 15, 16^,^^29^^,^^38^^,^^40^ while matched controls were all chosen from the same community as participants born MLP. GA was ascertained from medical records in seven cohort studies,14, 15, 16^,^^37^^,^^38^^,^^40^^,^^42^ while one study did not provide information on how preterm birth status was determined.^29^ For all cohort studies, the pulmonary function of participants was not known upon study commencement.14, 15, 16^,^^29^^,^^37^^,^^38^^,^^40^^,^^42^ Respective cohorts were age-matched in all studies. Most cohort studies (n = 5, 63%) controlled for an additional factor associated with pulmonary function in either study design or analysis.14, 15, 16^,^^38^^,^^42^ Outcome assessment was poor; one study reported blinding of assessor to the gestational age of participants,^14^ while another reported data collection by an independent assessor.^40^ A follow-up duration of five years or more was reported in all cohort studies. Only one study reported a follow-up rate of ≥70% of the eligible cohort.^29^ The case-control study achieved nine stars on the NOS for case control studies.^32^
Quality assessment of eight cross-sectional studies using the JBI appraisal checklist ranged from five to seven, with a median score of six. For all cross-sectional studies, the inclusion criteria, participants and setting were clearly described.^28^^,^^30^^,^^31^^,^^33^^,^^34^^,^^36^^,^^39^^,^^41^ Most cross-sectional studies (n = 7, 88%) ascertained preterm birth status by medical chart review.^28^^,^^30^^,^^31^^,^^33^^,^^34^^,^^39^^,^^41^ However, for all cross-sectional studies it was unclear if GA was measured using standard criteria, for example by ultrasound or parental reporting. Confounding factors associated with pulmonary function were identified for all cross-sectional studies, although only three studies implemented strategies to minimise the effect of these factors.^31^^,^^34^^,^^36^ All cross-sectional studies measured pulmonary function in a valid and reliable way and used appropriate statistical analysis. The RCT was appraised as ‘low risk’ using the Cochrane risk-of-bias tool for randomised trials.^35^ Quality assessment results for studies can be found in e-Appendix #3.
Compared with controls, participants born MLP had lower z-scores (mean difference [95% CI] I^2^) for FEV1: -0.22 [-0.35, -0.09] 49.3% (Figure 2), FVC: -0.23 [-0.40, -0.06] 71.8% (Figure 3), FEV1/FVC: -0.11 [-0.20 to -0.03] 9.3% and FEF25-75%: -0.27 [-0.41 to -0.12] 21.9%. Participants born MLP also had lower z-scores, on average, when compared with population norms (mean [95% CI] I^2^) for FEV1: -0.26 [-0.40 to -0.11] 85.2% (Figure 4), FVC: -0.18 [-0.34 to -0.02] 88.29%, FEV1/FVC: -0.24 [-0.43 to -0.05] 90.5% and FEF25-75%: -0.33 [-0.54 to -0.20] 94.8%. Supplementary forest plots are presented in e-Appendix #4. Similar results were also found when sensitivity analysis was conducted, excluding one study that did not present z-scores in GLI (e-Appendix #5). A sub-group analysis of those ≤ 18 years of age and studies of high quality only rendered similar overall estimates for all indices, except FEV1/FVC (e-Appendix #6). Egger's test for a regression intercept produced a p-value 0.043, indicating potential publication bias (e-Appendix #7).
Figure 2 Forced expiratory volume in 1 second z-score (zFEV
1) of the moderate-late preterm (MLP) group compared with the term group. A mean zFEV1less than 0 indicates individuals born MLP are performing worse than the population of term-born controls, on average.
Figure 3 Forced vital capacity z-score (zFVC) of the moderate-late preterm (MLP) group compared with the term group. A mean zFVC less than 0 indicates individuals born MLP are performing worse than the population of term-born controls, on average.
Figure 4 Forced expiratory volume in 1 second z-score (zFEV1) of the moderate-late preterm group compared with a z-score of 0. A mean zFEV
1less than 0 indicates individuals born MLP are performing worse than population norms, as derived from relevant reference equations, on average.
This study provides compelling evidence that children and adults born MLP have worse expiratory flows than individuals born at term. A reduction across all spirometric indices was observed, indicating an impairment in pulmonary function. However, the degree of airflow obstruction was modest, evident by mean differences in z-scores of -0.22, -0.11 and -0.27 for FEV1, FEV1/FVC and FEF25-75%, respectively, when compared with term-born controls. Similar results were found when comparing spirometric indices with expected population mean z-scores of 0. Reductions in z-scores of these indices suggest that those born MLP are failing to catch up to normal physiological levels and therefore peak function.
Todisco et al published a study investigating pulmonary function in children born MLP in 1993.^30^ The group performed spirometry and body plethysmography on 34 children born MLP, comparing them to matched term-born siblings. Despite similar expiratory airflows between the two groups, MLP children were found to have elevated residual volumes and residual volume to total lung capacity ratios. In 2012 and 2016, longitudinal findings from respective Welsh and Swedish birth cohorts were published. A modest reduction of expiratory airflows in 8-9-year-old children born MLP was observed in both cohorts, however there are mixed results on whether this trend persisted to the ages of 16-17 years, with the Welsh cohort demonstrating an almost complete catch up to term-born controls,^14^ while the Swedish cohort did not.^15^ Despite recent interest in the long term pulmonary function of MLP infants, a very limited number of studies to date have investigated measurements of pulmonary function other than expiratory airflows. Two studies have conducted multiple breath washout in conjunction with spirometry in school-aged children and adolescent groups, respectively. Lung clearance index estimates were similar between participants born MLP and term-born controls.^40^^,^^41^ Further to these findings, one of these studies also documented similar levels of exercise capacity between the two groups.^40^
Deficits in pulmonary function are more severe in children and adults born EP and VP than those born MLP. As such, the associations between gestational ages less than 32 weeks and reduced pulmonary function later in life are now well recognised. In 2012, findings from a meta-analysis provided evidence of an association between preterm birth and a reduced FEV1 later in life. The meta-analysis included summary data from 22 studies, encompassing a total of 2085 and 3820 preterm and term-born controls, respectively.^2^ In addition to the marked reduction in FEV1, those born preterm who did not develop bronchopulmonary dysplasia (BPD) had more favourable FEV1 outcomes later in life compared with those who did; an important observation given that BPD is uncommon among MLP newborns.^43^ Importantly, these 22 studies collated data from a wide range of gestational ages (23 - 36 weeks). Consequently, it is difficult to determine the contribution of the MLP group to the overall findings of that study.
Preterm birth may increase the risk of developing chronic obstructive pulmonary disease (COPD) later in life, with EP and VP birth status now considered a potential risk factor.^44^^,^^45^ A IPD meta-analysis published in 2019 demonstrated that VP survivors in early adulthood experience a mean reduction of -0.78 (95% CI -0.96, -0.61) in the z-score of FEV1 compared with term-born controls.^44^ Moreover, recent evidence indicates that middle-aged adults born moderate preterm during the pre-surfactant era experience impaired pulmonary function and therefore are at an increased risk of developing COPD compared with age-matched adults born late preterm and term.^45^ While the risk of later life COPD is substantially lower compared with VP survivors, those born MLP should be made aware of the potential consequences preterm birth may pose to long-term respiratory health. Personal smoking is likely to compound the effect of moderate preterm birth on pulmonary function,^45^ and therefore smoking avoidance in this group should be encouraged. MLP infants are more likely to experience adverse sequelae from viral infections.^46^^,^^47^ Subsequently, a lower threshold to seek medical care for respiratory symptoms may wish to be considered. Furthermore, mothers at high risk of preterm birth should be informed of the damaging effects that prenatal smoke and air pollutant exposure may have on postnatal pulmonary function.48, 49, 50
Given the potential long term consequences of MLP birth on respiratory health, it is essential to understand the economic and healthcare burden that rising rates of MLP birth may cause. A recent decision-analytic model estimated that for the first 18 years of life, a hypothetical birth cohort of 314,814 children - the number of total births in 2016 in Australia – a preterm birth rate of 8.5% will bear a cost of 1.413 billion Australian dollars, 39% of which is attributed to those born MLP.^7^ A similar societal economic burden also exists within the US, where the current total incremental lifetime cost of an infant born MLP in 2016 is estimated to be 28,367 US dollars.^51^ A 2016 MLP birth rate of 8.26% equated to 325,361 neonates being born MLP in the US.^52^ Consequently, the incremental lifelong cost of this birth cohort is estimated to reach 9.2 billion US dollars.^51^ In addition to lower expiratory airflows, emerging data suggest that MLP birth is associated with increased cardiometabolic risk and impaired neurodevelopment and social-emotional development compared with term birth.^53^^,^^54^ Poorer outcomes across multiple health domains in conjunction with continued improvements in preterm birth survival rates is likely to lead to a further burden on healthcare systems. To address changes in the epidemiology of preterm birth alongside increasing preterm survival rates, high-quality studies investigating long-term health outcomes of preterm survivors are urgently required to inform government policy and clinical guidelines.^55^
This review provides an update and significant contribution to the current evidence base of the effect of MLP birth on later life expiratory airflows, with all studies eligible for meta-analysis having been conducted in the last decade. A major strength of our meta-analysis is that outcomes expressed as GLI z-score were obtained for all but one study, reducing a potential source of heterogeneity and making these results more relevant to current clinical practice. On the contrary, while all but one study reported worse expiratory airflows in those born MLP when compared with term-born controls, high levels of heterogeneity were observed in several of the analyses. Moreover, evidence of asymmetry due to heterogeneity was found when a funnel plot analysis and Egger's test were performed on studies that reported an FEV1 z-score from participants born MLP. Consequently, caution is advised when interpreting these findings. In addition, comparing to a population mean of 0 may have introduced bias given that poor GLI fit has been previously identified among several population groups.56, 57, 58 We acknowledge that a limitation of our review is that we were unable to investigate longitudinal changes in expiratory airflows between MLP participants and term-born controls. Of the three studies with longitudinal outcomes, one used different reference equations at each time point,^37^ while two studies reported z-scores derived from different reference equations.^14^^,^^15^
Due to a lack of longitudinal data, in conjunction with continuing improvements in neonatal care and reductions in the prevalence of smoking during pregnancy,^59^ secular trends in the long-term pulmonary function of future MLP birth cohorts are difficult to determine. It remains unclear if MLP birth affects gas diffusion, static lung volume or lung clearance index. Future studies may wish to consider assessing pulmonary function measurements other than baseline spirometry to fully understand the effects of MLP birth on later life respiratory health. Finally, associations between perinatal and postnatal factors, such as early childhood respiratory illness, and pulmonary function later in life are poorly described in MLP cohorts. Further research is required to discern if reductions in later life pulmonary function are a result solely of preterm birth itself, or in conjunction with early life factors associated with MLP birth.
In conclusion, this study provides evidence that children and adults born MLP experience worse expiratory airflows than those born at term, although the reductions are small. While children and adults born MLP have more favourable long term respiratory outcomes than EP and VP survivors, general practitioners and pulmonary specialists should be aware that those born MLP may be at higher risk of COPD compared with term-born peers.
CDB, CN, JLYC, TF, RM, SR, LWD, EJLEV and LW conceived and designed the study. CDB and CN performed the literature search. CDB, CN, JLYC, TF, RM, SR, LWD, EJLEV and LW were involved in data collection and interpretation. CDB, RM, and LW were involved in data analysis. CDB, CN, JLYC, TF, RM and LW were involved in data interpretation. CDB drafted the manuscript. All authors were involved in revising the manuscript and approved the final submitted version. CDB, CN and LW accessed and verified the underlying data reported in the manuscript.
All data used and generated in this study are available within the article and/or in the supplement material.
We declare no competing interests.
We wish to acknowledge the following individuals for their contribution of aggregate-level data for this meta-analysis: J. Hallberg (Children and Youth Hospital, Södersjukhuset), J. Landry (McGill University, Montreal), J. Michael Collaco (Johns Hopkins University School of Medicine, Baltimore), K. Korsten (University Medical Center Utrecht, The Netherlands), L. Bont (University Medical Center Utrecht), P. Näsänen-Gilmore (National Institute for Health and Welfare, Helsinki and Oulu), Y. Salem (Bern University Hospital, University of Bern) and X. Carbonell-Estrany (Obstetricia Neonatologia, Barcelona). This work is supported by grants from the National Health and Medical Research Council (Centre of Research Excellence #1153176, Project grant #1161304); Medical Research Future Fund (Career Development Fellowship to J.L.Y Cheong #1141354) and from the Victorian Government's Operational Infrastructure Support Programme. C. Du Berry's PhD candidature is supported by the Melbourne Research Scholarship and the Centre of Research Excellence in Newborn Medicine.