Outcomes of postnatal systemic corticosteroids administration in ventilated preterm newborns: a systematic review of randomized controlled trials

Introduction Prolonged mechanical ventilation, commonly used to assist preterm newborns, increases the risk of developing bronchopulmonary dysplasia (BPD). In recent decades, studies have demonstrated that systemic corticosteroids play a significant role in the prevention and management of BPD. In this systematic review of randomized controlled trials (RCTs), we evaluated the association between the administration of systemic corticosteroids in preterm infants and its long-term outcomes, such as neurodevelopment, growth, extubation rate, and related adverse effects. Methods We conducted an electronic search in Medline, Scopus, and PubMed using the following terms: “premature infants” and “corticosteroids.” We considered all RCTs published up to June 2023 as eligible. We included all studies involving preterm newborns treated with systemic corticosteroids and excluded studies on inhaled corticosteroids. Results A total of 39 RCTs were evaluated. The influence of steroids administered systemically during the neonatal period on long-term neurological outcomes remains unknown, with no influence observed for long-term growth. The postnatal administration of systemic corticosteroids has been found to reduce the timing of extubation and improve respiratory outcomes. Dexamethasone appears to be more effective than hydrocortisone, despite causing a higher rate of systemic hypertension and hyperglycemia. However, in the majority of RCTs analyzed, there were no differences in the adverse effects related to postnatal corticosteroid administration. Conclusion Dexamethasone administered during the neonatal period appears to be more effective than hydrocortisone in terms of respiratory outcomes; however, caution should be taken when administering dexamethasone. Data derived from current evidence, including meta-analyses, are inconclusive on the long-term effects of the administration of systemic steroids in preterm infants or the possibility of neurodevelopmental consequences.


Introduction
The survival rate of preterm newborns has improved over the last 20 years due to advances in neonatal care (1).However, the improvement in survival has been associated with an increased morbidity rate and reduced long-term neurodevelopmental (NDV) outcomes (2,3).Two of the most important innovations in neonatal care are the introduction of surfactant therapy and the improvement in mechanical ventilation (MV).However, prolonged MV (PMV) is harmful and increases the risk of developing bronchopulmonary dysplasia (BPD) (4,5).BPD is a chronic inflammatory lung disease of premature neonates characterized by impaired lung development (4,6).It has a multifactorial pathogenesis, wherein prolonged oxygen exposure induces a destructive local inflammatory response in the lung alveoli, associated with a simultaneous impaired repair response (6).In addition, it has been demonstrated that BPD is associated with impaired long-term NDV and pulmonary function outcomes (7,8).Thus, neonatologists aim to extubate preterms as soon as possible, albeit not always possible, especially for extremely low gestational age newborns (ELGAN).Corticosteroids (e.g., dexamethasone and hydrocortisone) are currently administered intravenously (IV) or orally (PO) for the treatment and prevention of BPD.Studies have related their beneficial effects to their antiinflammatory activity (9)(10)(11).The authors researched inhaled corticosteroids (12); however, no beneficial effects were found on the risk of neurological disability although the mortality rate was higher in the treated group, thus not allowing the routine administration (13,14).However, whether the use of dexamethasone or hydrocortisone IV or PO improves or reduces long-term neurological outcomes is still debated.In addition, there are concerns regarding the prophylactic use of systemic corticosteroid therapy for possible adverse effects (i.e., sepsis, infection, and metabolic side effects).
In this systematic review of randomized controlled trials (RCTs), we studied the association between the administration of systemic corticosteroids during the neonatal period and its longterm outcomes, in terms of NDV and growth.Additionally, we evaluated the respiratory outcomes and possible adverse effects.

Studies, population, and intervention
We considered all RCTs published up to June 2023 as eligible.We included all studies involving preterm newborns treated with systemic (IV or PO) corticosteroids and excluded studies on inhaled corticosteroids.

Outcomes
Our primary outcome was the long-term effects such as NDV and growth.Our secondary outcomes were as follows: the rate of extubation, reintubation, BPD, other respiratory outcomes (considering the duration of invasive or non-invasive MV, supplemental oxygen therapy, FiO2, or other specific ventilatory/ respiratory data), and steroid-related adverse effects.We considered the following as adverse effects: systemic hypertension, hyperglycemia (HG), sepsis or other infections, patent ductus arteriosus (PDA), intraventricular hemorrhage (IVH), necrotizing enterocolitis (NEC), and retinopathy of prematurity (ROP).

Research methods and study selection
We performed a systematic review of the published RCTs, in compliance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines (15).We conducted an electronic search in Medline, Scopus, and PubMed using the following medical subject headings and terms: "premature infants" and "corticosteroids."Only English manuscripts and RCTs were considered.Two authors (G.B. and F.L.) independently assessed the study eligibility according to the pre-established criteria and performed an accurate check to exclude duplicates.A discussion with a third part researcher (G.T.) resolved different in opinion, to achieve consensus.We performed a manual search of the reference list of the systematic reviews and meta-analyses published and excluded them from this review.
Data extraction, management, and risk of bias doses of steroids, days to extubation, duration of therapy, more than one cycle of steroids, and administration of other steroids.Another specific data form summarized the outcomes (e.g., extubation, reintubation, BPD, other respiratory outcomes, systemic hypertension, HG, sepsis/infection, PDA, IVH, NEC, ROP, and long-term outcomes).These data were checked for missing information, errors, and inconsistencies with published reports.If evidenced, differences were resolved by discussion and consensus between the researchers.The corresponding authors were contacted when the eligibility criteria of their papers were unclear.
The risk of bias was assessed independently by two researchers (G.B. and G.T.) using a specific form.We considered bias as selection bias (random sequence generation and allocation concealment), performance bias (blinding of the study personnel as to which intervention a neonate had received), detection bias (blinding of personnel evaluating outcomes), attrition bias (completeness of reporting data, reason, and balance across groups of missing data), reporting bias (reporting of the study's prespecified or expected outcomes of interest to the review), and other sources of bias (early interruption of the trial due to data-dependent process or bias related to the specific study design).We categorized the risks of bias as high, low, or unclear for each study, using standard methods (15).The selection bias was judged as unclear when these aspects were not available.The differences in opinion were resolved by discussion and consensus.

Study description
During the research process using the mesh term described in the Materials and methods section, we found 1,011 articles, and 11 RCTs were selected in the first qualitative synthesis after the screening process (Figure 1).After a manual search of the reference list of the systematic reviews and metaanalyses analyzed in the previous stages, we added 28 RCTs, and 39 RCTs were analyzed in the final step of this systematic review (Figure 1).Data extracted are summarized in Tables 1-6 .

Primary outcome: long-term NDV effects and growth
A graphical representation of the percentage of the studies evaluating the outcomes of interest for this study is shown in Figure 2  .
Eight studies did not find statistically significant long-term neurological differences (17,21,23,31,35,48,50,52). Malloy et al. (49) found an increased risk of NDV delay, and O'Shea et al. (37) showed an increased risk of cerebral palsy at 1 year in the groups receiving dexamethasone (different doses vs. placebo, respectively).A long-term follow-up study performed by Yeh et al. (33) 19) found better NDV outcomes at 2 years in newborns who received hydrocortisone if they were born at 24-25 weeks of GA, while no statistical difference considering those born at 26-27 weeks of GA.All the RCTs that considered growth parameters did not find differences between the two groups (17,23,24,31,35,37,46,47,52).Only Yeh et al. (33) found that newborns who received dexamethasone were significantly shorter than the controls and had a significantly smaller head circumference, evaluated at school age.However, the long-term NDV and growth outcomes evaluated in these 15 RCTs were measured at different time points (Tables 4-6).
To better characterize the effects of corticosteroids on longterm neurological outcomes, we performed also a sub-analysis separating the studies for early (before 7 days of life) and late (after 7 days of life) administration.The graphical representation of this sub-analysis is reported in Supplementary Figure S1.We excluded the studies of Baud et al. (19,20) for this analysis, because the starting age of the intervention was before 10 days of life, and based on the definition of early and late administration of corticosteroids considered, we were not able to add these studies in the analysis.

Risk of bias
The quality of the studies was assessed by risks of bias, as shown in Figure 3.We judged the risk of selection bias as low in all uncontrolled studies, unclear for 14, and high for 2 RCTs.Performance bias and detection bias were high for four and three studies, respectively, and unclear for one and five, respectively.The rest of the 34 and 31 RCTs were judged as low risk of performance and detection bias, respectively.Attrition bias was judged as low for 29 studies, high for 2, and unclear for 8 RCTs.Considering our outcome, reporting bias was judged high for 8, low for 29, and unclear for the remaining 2 RCTs.Other sources of bias were not evaluable for 6 controlled trials, while was high for 23 studies and low only for 10.

Discussion
Despite many studies exploring the effects of corticosteroid administration in postnatal life for preterm newborns, the optimal modality of administration remains to be defined.The high heterogeneity of the included RCTs regarding dosages, timing of administration, and outcome measures, discourages the mathematical analysis of the data (55).A recent meta-analysis and network meta-analysis included only one study for a therapeutic regiment or more than one study but with different timing of administration and were focused primarily on BPD (11).In this systematic review, we evaluated the efficacy of steroid therapy on other respiratory outcomes, including extubation rate, reintubation, BPD, and related side effects, considering separately hydrocortisone, dexamethasone, and different doses of dexamethasone.Despite the analyzed studies suggesting an overall favorable effect of steroids on respiratory outcomes, a well-designed large RCT is urgently needed to establish the optimal indication and modalities of administration.Among the steroids used for preterm newborns, early and high doses of dexamethasone have a better impact on the respiratory outcome, while hydrocortisone is related to fewer side effects such as systematic hypertension or HG.

Long-term NDV effects and growth
A major limitation in the analysis of the studies including the use of steroids in the neonatal period is represented by the lack of data for long-term outcomes.Only a few studies evaluated neurological and growth outcomes of preterm newborns receiving steroids in early life (17,19,21,23,24,31,35,37,(46)(47)(48)(49)(50)52).Only one-third of the trials on hydrocortisone evaluated the long-term effects on NDV up to 2 years of life (17,19,21,23,24).In these studies, neurological outcomes seem to be not influenced by the use of hydrocortisone.Two RCTs found an improvement in neurological outcomes, particularly in ELGAN and newborns with extremely low birth weight (19,24).Almost half of studies on dexamethasone evaluated neurological outcomes up to 1-2 years of life (31,35,37,(46)(47)(48)(49)(50)52).The results of these studies are controversial.If the majority of the RCTs demonstrated that there is no influence of dexamethasone on long-term NDV (31,35,48,50,52), O'Shea et al. (37) reported an increased rate of cerebral palsy at 1 year for newborns treated with dexamethasone (42 days of therapy, starting doses at 0.25 mg/kg twice a day for 3 days) compared with placebo.Malloy et al. (49) found a worse NDV long-term outcome for newborns in the high doses of dexamethasone groups (0.5 mg/kg/day for 3 days followed by 0.3 mg/kg/day for 4 days every 12 h vs. 0.08 mg/kg/day for 7 days every 12 h).In addition, Yeh et al. (33) showed that children treated with two doses per day of dexamethasone for 28 days (0.25 mg/kg/dose up to 7 days and then the dose was tapered) had significantly poorer motor skills, motor coordination, visual-motor integration, and significantly lower full IQ, verbal IQ, and performance IQ scores.
The frequency of clinically significant disabilities was significantly higher among children in the dexamethasone group compared with controls (39% vs. 22%, p value 0.04) (33).However, O'Shea et al. did not power the study for long-term NDV and selection bias because differential survival rates across the two study groups could explain the greater risk of cerebral palsy among the intervention group; Malloy et al. performed a trial with a very low sample size (8 vs. 8).On the other hand, Cummings et al. and Marr et al. found a better long-term neurological outcome for babies treated with dexamethasone (42 days of therapy, starting dose at 0.5 mg/kg/day for the first 3 days, 0.3 mg/kg/day for the next 3 days, and then reduced by 10% every 3 days until a dose of 0.1 mg/kg was reached at day 34) (46,47).They enrolled newborns born ≤30 weeks of GA (and ≤1,250 g) and ≤27 weeks of GA, respectively.Thus, it could be possible that ELGAN should benefit more than other newborns from steroid treatment.In addition, Marr et al. performed a study with a long time follow-up evaluation of 7 years.Most of the studies did not consider long-term growth as an outcome.The available studies suggest a low impact of early steroid treatment and long-term growth (17,23,24,31,35,37,46,47,52).Only Yeh et al. demonstrated that dexamethasone could negatively influence height and head circumference, evaluated at school age (33).

Respiratory outcome, metabolic effects, and morbidity during hospital stay
Postnatal steroid treatment is beneficial for respiratory outcomes, such as extubation, reduced risk of BPD or duration of invasive or non-invasive MV, supplemental oxygen therapy, FiO2, and/or other specific ventilatory/respiratory data.Dexamethasone has a better impact compared to hydrocortisone, but the optimal therapeutic regiment remains to be defined.For the reintubation rate outcome, a conclusion cannot be made, since it has been evaluated only in a few, unpowered studies (32,(45)(46)(47).In addition, in all these studies, dexamethasone was used as an intervention, while none of them considered hydrocortisone.
Our analysis showed that BPD is the most studied respiratory outcome.The majority of the studies demonstrated that dexamethasone had a positive effect on the BPD rate, whereas hydrocortisone appears to not improve this outcome.The study with a lower risk of bias showed that the therapeutic regiment of dexamethasone in early life (0.5 mg/kg/day for the first 3 days, 0.25 mg/kg/day for the next 3 days, and 0.125 mg/kg/day on the day 7) had the best impact on BPD (36).
Steroids work as anti-inflammatory agents, which can explain their rationale in the prevention of BPD.One of the main risk factors for developing BPD is prolonged oxygen exposure and MV, which induce a pulmonary local inflammatory response (4,5).Postnatal corticosteroids decrease inflammation and edema, improving gas exchange and lung protective mechanisms (56, 57).
To the best of our knowledge, there are no RCTs that evaluated the long-term pulmonary outcome of preterm newborns treated with corticosteroids in neonatal life.On the other hand, studies aimed at investigating long-term respiratory function in preterm babies were mainly focused on BPD and did not independently evaluate the role of postnatal steroid treatment on the final outcome (58).
Not all studies evaluated side effects associated with the use of steroids in preterm newborns.
Systemic hypertension was analyzed in the majority of the studies included in this systematic review.Few studies (∼16%) reported an increased risk of hypertension in newborns treated with steroids.The administration of dexamethasone has been associated with systemic hypertension more frequently compared to hydrocortisone.Treatment regimens of more than 0.25 mg/kg of dexamethasone, especially for more than 10 days, appear to increase the risk of systemic hypertension.
HG is an independent risk factor for mortality and NDV delay in newborns (59, 60); thus, all efforts should be made to reduce the risk of this condition.A higher number of studies on dexamethasone reported an increased risk of HG compared with studies on hydrocortisone.However, all the studies that compared two different doses or timing for dexamethasone administration found no difference for HG (47)(48)(49)(50)(51)(52)(53)(54).When steroid treatment is needed, neonatologists should minimize the other conditions that induce an increased risk of HG (such as nutrition) (60,61), or they should improve continuous glucose monitoring to maintain euglycemia (62).
Sepsis remains the major cause of morbidity and mortality in preterm newborns (63).Our analysis showed a relevant increased risk of sepsis in newborns treated with steroids, more for hydrocortisone compared with dexamethasone.However, a recent study demonstrated a reduced risk of pneumonia (18) in newborns treated with hydrocortisone, probably due to an improvement in respiratory outcome and a reduced time of invasive ventilation support.
Data regarding the relationship between steroid administration and morbidity conditions are controversial.Most studies suggested no relation (about 69%), while others demonstrated an increased rate of prematurity-related morbidities (about 14%); finally, about 10% suggested a better outcome.We speculated that this result depends on the different modalities of administration of corticosteroids, different therapeutic regiments, and different morbidity definitions of the studies.Based on our findings, more than 42 days of dexamethasone therapy, more than 1 mg/kg starting dose, or more than 10 days of hydrocortisone therapy might increase the risk of morbidity.
We speculate that part of the reduction in morbidity might be related to the improvement in PDA closure, associated with steroid use.Some studies demonstrated that steroids (especially hydrocortisone) could improve the PDA outcome, reducing the need for medical or surgical treatment (19,20,23,27,40).Several reasons may explain this effect: (1) in vitro studies demonstrated that hydrocortisone treatment decreases the sensitivity of the ductus arteriosus to the relaxing action of prostaglandin E2, which explains the beneficial effects in vivo of steroids (64,65), and (2) the relationship between PDA and BPD, especially for ELGAN, has been demonstrated (66).Despite this topic is not well evaluated, it could be possible that the effects of steroid administration on PDA also improve BPD.However, future studies should evaluate this aspect.
Brozanski et al. (42) found a reduced risk of IVH for preterm receiving steroids.The authors administered a pulse dose of 0.25 mg/kg/dose of dexamethasone to newborns at 7 days of life for 3 days, repeated every 10 days until 36 weeks of postmenstrual age or up to weaning of ventilation/oxygen support.In addition, they affirmed that the decreased rate of IVH in the intervention (pulse) group could be associated with a better stabilization of capillary membranes or alteration of cerebral blood flow by corticosteroids or by an improvement in the ventilatory status of the infants (42).However, two RCTs underline the effects of steroids on early brain damage (37,38).They administered 0.25 mg/kg twice a day for 3 days, 0.15 mg/kg twice a day for other 3 days, and then a 10% reduction in the dose every 3 days until the dose of 0.1 mg/kg was reached on day 34 and after 3 days on this dose.However, all three studies are not powered for this outcome.Thus, considering the different therapeutic regiments and this limitation, further studies are needed to clarify this aspect.

Risk of bias
Given that some studies with concerns about the overall risk of bias have been included, our results need to be confirmed by further RCTs with a low risk of bias.Blinding bias was low for most of the studies analyzed.Attrition and reporting bias were judged low for most of the studies reviewed.Major concerns are about the risk of other bias (early interruption of the trial due to data-dependent process or bias related to the specific study design), which was judged high for 23 of 38 studies.In addition, the studies presented a high heterogeneity of inclusion criteria (e.g., GA and/or BW) and intervention (different doses, timing of administration, and duration therapy), which could have influenced the results.We included only RCTs, despite some of these enrolling a small number of patients with low power of the study.Some of the studies included adopted a non-optimal blinding method or were unclear.

Strengths and limitations
Our results should be interpreted considering the limitations of the studies analyzed and of the review process.First, we decided not to perform a meta-analysis because of the extreme variability in methodology, modality of administration of steroids, and outcome of the studies (55).We systematically collected evidence and after a deep evaluation and discussion between the authors, we decided not to make a meta-analysis considering the wide differences in methodology used in different studies included in this manuscript.In particular, the studies vary regarding inclusion and exclusion criteria, enrollment, dose of treatment, starting days and duration of steroids, type of steroids administered, timing of follow-up, and assessment scales (Tables 1-3).The data deriving from current evidence, including meta-analyses, are inconclusive on the long-term effects either to exclude completely that there may be consequences on the central nervous system.Thus, we believe that steroids should be used in trial settings and to collect data in large databases to verify the consequences of this therapy.
We believe that there are no minimal criteria to perform a meta-analysis and that conclusions deriving from published meta-analyses were not supported by robust statistical data.Our data might contribute to better define the modality of steroid therapy and the target population to reduce the risk of brain damage.Whether meta-analyses suggested deleterious effects of steroids on NDV, our study demonstrated that further welldesigned studies are needed to reach conclusions regarding the relationship between steroid treatment in preterm newborns and long-term NDV.
We synthesized the results of different studies on NDV.However, the long-term outcomes were not analyzed at the same time point and with different NDV assessment scales.We selected articles published in the English language; thus, it is possible that some gray literature has not been analyzed.In addition, studies showing positive results have a greater likelihood of being published.Finally, in some studies, other medications were administered, based on the clinical conditions of patients, in some studies treatment was interrupted prematurely, and others were not powered for the long-term outcomes.

Conclusion
Postnatal administration of systemic corticosteroids is an important tool for neonatologists to improve respiratory outcomes.Based on published RCTs, dexamethasone appears to be more effective than hydrocortisone for extubation, prevention of BPD, and improvement of respiratory outcomes.However, considering the deleterious effects such as HG, caution should be made during administration of dexamethasone.In addition, long-term effects on NDV and growth remain undefined.Considering that data deriving from current evidence, including meta-analyses, are inconclusive on the long-term effects to exclude completely that there may be consequences on the central nervous system, further studies are advocated to define the optimal therapeutic regiment, to improve the positive effects and reduce the side effects of steroid administration in preterm newborns.

FIGURE 3
FIGURE 3Risk of bias of the evaluated randomized controlled trials.

TABLE 1
Randomized controlled trials comparing hydrocortisone vs. placebo.

TABLE 4
Randomized controlled trials comparing hydrocortisone vs. placebo.
showed worse effects on neuromotor and cognitive function at school age in newborns who received dexamethasone compared to the placebo group.Watterberg et al. (24) (hydrocortisone vs. placebo), Cummings et al. (46) (dexamethasone vs. placebo), and Marr et al. (47) (different doses of dexamethasone) found an improved long-term NDV.Baud et al. (

TABLE 6
Randomized controlled trials comparing different doses of dexamethasone.