Results of the search

For the results of our literature searches see Study flow diagram (Figure 1).

Figure 1

---

---

Study flow diagram.

For this update of the review two additional trials were identified (Bassler 2015; Nakamura 2016). The study by Denjean 1998 was moved from excluded to included studies.

Fourteen trials assessing the impact of inhaled corticosteroids were identified, of which four trials were excluded (Kovacs 1998; Beresford 2002; Dugas 2005; Yeh 2016).

Ten trials qualified for inclusion in this review: Denjean 1998; Townsend 1998; Yong 1999; Merz 1999; Fok 1999; Cole 1999; Jonsson 2000; Jangaard 2002; Bassler 2015; and Nakamura 2016. Eight included studies have been published as complete articles while two studies were presented at scientific meetings and published in abstract form (Townsend 1998; Yong 1999). Complete data from the investigators were available for Yong 1999, while information published in abstract form is presented for Townsend 1998.

Although all studies attempted to include infants thought to be at risk of developing CLD, the inclusion criteria, the intervention (type of inhaled corticosteroid) and duration of therapy varied between studies. All studies except Jonsson 2000 used a metered dose inhaler (MDI) and an Aerochamber interposed between the endotracheal tube and the ventilatory circuit or a manual puffer (anaesthesia bag). In Cole 1999, the drug was administered to infants who were extubated using the same procedure through a nasopharyngeal tube. In Jonsson 2000, a dosimetric jet nebulizer was used to deliver the aerosol.

Details of each study are given in the table 'Characteristics of included studies'.

Included studies

See 'Characteristics of included studies'.

Ten trials qualified for inclusion in this review: Denjean 1998; Townsend 1998; Yong 1999; Merz 1999; Fok 1999; Cole 1999; Jonsson 2000; Jangaard 2002; Bassler 2015; and Nakamura 2016.

Bassler 2015: the study by Bassler and co‐workers enrolled 863 infants with postmenstrual age (PMA) 23^0/7 weeks to 27^6/7 weeks and chronological age of 12 hours or less, who required any form of positive pressure support. Exclusion criteria were: palliative care; dysmorphic feature or congenital abnormalities likely to affect life expectancy or neurologic development; strongly suspected cyanotic heart disease; and, to prevent a correlated data problem, the infant was from a multiple‐birth pregnancy (other than the second infant in birth order, who was considered eligible). The budesonide group (n = 441; 437 followed to first discharge home) received two puffs of budesonide (200 µg/puff) administered every 12 hours for the first 14 days of life and one puff administered every 12 hours from day 15 until the last dose of study drug had been administered. Study drugs were administered until infants no longer needed supplemental oxygen and positive pressure support or reached a PMA of 32^0/7 weeks, regardless of ventilator status. The control group (n = 422; 419 followed to first discharge home) received two puffs of placebo containing only hydrofluoroalkane propellant.

Primary outcome: a composite of death or bronchopulmonary dysplasia (BPD) at 36 weeks' PMA. BPD was defined as the requirement for positive pressure support, the requirement for supplemental oxygen at FiO₂ exceeding 0.30, or in infants receiving low amounts of oxygen, an inability to maintain an oxygen saturation value above 90% during a structured, short period of saturation monitoring coupled with gradual weaning from oxygen to ambient air (oxygen reduction test).

Secondary outcomes included: death by any cause at 36 weeks' PMA, BPD at 36 weeks' PMA (defined as per above); duration of positive pressure respiratory support or supplemental oxygen; ventriculomegaly with or without intraventricular haemorrhage (IVH) on ultrasound at or before 36 weeks' PMA; patent ductus arteriosus (PDA) requiring medical or surgical treatment; intestinal perforation or necrotizing enterocolitis (NEC); retinopathy of prematurity (ROP) (stage 2 or higher); culture‐proven infections; increase in body weight and head circumference from birth to day 28; length of hospital stay; need for reintubation after the last dose of drug had been administered; occurrence of oral candidiasis requiring treatment; hyperglycaemia requiring insulin treatment; and hypertension requiring treatment.

Neurodevelopmental disability testing is to be conducted at 18 to 22 months (hence is not reported in this current publication).

Cole 1999: the study by Cole and co‐workers enrolled infants of less than 33 weeks gestational age (GA) and with birth weight equal up to 1250 grams who required assisted ventilation at three to 14 days of life. The infants were randomly assigned to receive inhaled beclomethasone dipropionate (n = 123) or placebo (n = 130). The desired dose was calculated to deliver to the lung 40 µg/kg/day for the first week, 30 and 15 µg/kg/day for the second and third weeks respectively, and then 10 and 5 µg/kg/day during the fourth week. The primary outcome was the incidence of BPD in survivors defined as an abnormal chest x‐ray and need for supplemental oxygen at 28 days of life. Secondary outcomes included the incidence of BPD in survivors at 36 weeks' PMA (defined as an abnormal chest x‐ray and need for supplemental oxygen at that age), duration of respiratory support (oxygen, mechanical ventilation and continuous positive airway pressure), the need for systemic glucocorticoids, diuretic or bronchodilator therapy, death, length of hospitalisation and the incidence of complications possibly attributable to the use of inhaled steroids. Systemic glucocorticoid therapy was permitted at the discretion of the infant's physician. The baseline characteristics of the two groups were similar except for maternal race or ethnic group (P = 0.03). Two additional reports from the same trial (Cole 1999) have been published as separate articles. One of these reports evaluated the effect of inhaled beclomethasone therapy on adrenal response and the other report evaluated its effect on tracheal aspirate inflammatory mediators (IL‐8 and IL‐1ra).

Denjean 1998: the study by Denjean and co‐workers enrolled infants with respiratory distress syndrome and PMA less than 31 weeks. Infants were eligible for the study if they required ventilator support (intermittent mandatory ventilation (IMV) or nasal IMV/continuous positive airway pressure (CPAP)) on the 10^th postnatal day. 178 infants were randomised, five were withdrawn leaving 173 infants in the trial who were assigned to four groups (placebo + placebo; placebo + salbutamol; placebo + beclomethasone; beclomethasone + salbutamol) of which two (placebo + placebo and placebo + beclomethasone) are included in our review. Beclomethasone (250 µg) was given four times a day (1000 µg daily). Treatment was started on the 10^th or 11^th postnatal day and was given for 28 days, with dose tapering over a period of eight days. Total number reported on: n = 43. Placebo (250 µg) was given four times a day (1000 µg daily). Treatment was started on the 10^th or 11^th postnatal day and was given for 28 days, with dose tapering over a period of 8 days. Total number reported on: n = 43. Beclomethasone and placebo were administered by metered‐dose inhalers.

The main outcome criterion was CLD. The diagnosis of CLD was made at 28 days of age on the basis of clinical (oxygen dependence) and radiographic criteria. CLD was categorized in three grades of severity: severe: ventilation with endotracheal tube (ET) for more than three months or oxygen supplementation for more than four months; moderate: ventilation with ET for more than one month or oxygen supplementation for more than two months; and mild: ventilation with ET for less than one month and oxygen supplementation for less than two months. The definition of CLD was unclear. Secondary outcomes included: survival without CLD, death (during hospital stay), ventilatory support (nasal IMV, CPA or IMV) (days), need for supplementary oxygen (days), need for systemic dexamethasone (IV), and sepsis (positive blood culture).

Fok 1999: the study by Fok and co‐workers enrolled 53 infants born at less than 32 weeks' GA, birth weight less than 1.5 kg, requiring mechanical ventilation with an arterial PO₂/alveolar PO₂ ratio of less than 0.25 at 6 to 10 hours after the second dose of surfactant was administered. Infants were excluded if they needed high‐frequency ventilation at the time of enrolment. Infants were randomised to receive inhaled fluticasone propionate or placebo. Two puffs of fluticasone propionate (250 micrograms/puff) or placebo were administered 12 hourly for two weeks. The first dose was administered within 24 hours of birth. The primary outcome was successful extubation by day seven or day 14 of life. Secondary outcomes included mortality, oxygen dependency at 28 days of postnatal age and at 36 weeks' PMA. The incidence of complications possibly attributable to the use of inhaled steroids was monitored. Twenty‐seven infants were enrolled in the fluticasone propionate group and 26 in the placebo group. The two groups were similar in their demographic and perinatal characteristics.

Jangaard 2002: the study by Jangaard and co‐workers enrolled 60 preterm infants weighing less than 1250 grams with respiratory distress syndrome (RDS) and requiring ventilatory support at 72 hours of age. Infants were randomly assigned to receive inhaled beclomethasone dipropionate (250 µg/puff) or placebo for four weeks. Medication dosage assumed a 10% deposition of the administered dose with the aim to provide a total dose of 0.2 mg/kg/day. When the infants were extubated, the study drug was administered using an infant‐sized Aerochamber (Boehringer Ingelheim, Canada) with an appropriately fitted mask. The primary outcome for this study was BPD defined as oxygen dependency at 28 days of life. The demographic characteristics of the two groups were similar. Jangaard 2002 has been previously published in abstract form. The results of the published report are presented in this review.

Jonsson 2000: The study by Jonsson and co‐workers randomised 30 VLBW infants with median (range) GA of 26 weeks (23 to 29 weeks) and median birth weight of 805 g (525 to 1227 g). Inclusion criteria were 1) mechanical ventilation on day six of life, or 2) if extubated, nasal continuous positive airway pressure with FiO₂ of 0.3 or higher. Infants with the following conditions were excluded: congenital malformations, congenital heart disease and grades III‐IV IVH. Infants on high frequency oscillatory ventilation (HFOV) were excluded as the inhalations could not be given through the electronic dosimetric jet nebulizer. Infants were randomised to receive 500 µg twice a day or placebo delivered using a dosimetric jet nebulizer with variable inspiratory time and breath sensitivity. Inhalations were started on day seven of life. The primary objective was to attain a 30% reduction in FiO₂ levels in the budesonide treatment group after 14 days of therapy. Secondary outcomes included: duration of supplemental oxygen, duration of mechanical ventilation, duration of nasal CPAP, oxygen requirements at 28 days of age and at 36 weeks' postmenstrual age. Adrenal cortisol response to stimulation was measured at baseline (prior to commencement of inhalation) and at the end of the study period. Information on adverse events — hyperglycaemia, hypertension, sepsis, PDA, IVH and gastrointestinal problems — were collected. Of the 30 infants enrolled, one parent declined to participate and two eligible infants could not be included due to ongoing HFOV on day seven of life. Thirteen infants were enrolled in the budesonide group, of which eight were ventilated, while 14 infants were enrolled in the placebo group, of which nine were ventilated at the commencement of therapy. Only one outcome for ventilated infants was reported (successful extubation during the study period ‒ 14 days).

Merz 1999: the study by Merz and co‐workers enrolled 24 infants with a birth weight of 750 to 1500 grams, GA of 25 to 32 weeks, ventilator dependency on day three of life with a rate of 15 breaths/minute or more and FiO₂ of 0.25 or more to maintain oxygen saturation 90% or more. Infants were randomly assigned to inhaled budesonide (200 µg/puff) or placebo. Two puffs were administered four times a day for a total of 10 days. The primary outcome was duration of artificial ventilation. Secondary outcomes included the duration of supplemental oxygen and the release of albumin and different inflammatory mediators in the tracheobronchial aspirate fluid. Adverse events such as frequency of acute infections, hypertension, hyperglycaemia, and adrenal suppression were evaluated. The demographic and perinatal data were similar in both groups on the day of randomisation.

Nakamura 2016: The study by Nakamura and co‐workers enrolled 211 infants with birth weight of less than 1000 grams, who needed endotracheal intubation and respiratory support due to respiratory failure. Infants were randomised to prophylactic inhaled steroids starting within 24 hours of birth and continuing until six weeks of age or extubation. Two doses of 50 µg fluticasone propionate (FP) were administered every 24 hours. The placebo group received two doses of placebo every 24 hours. The placebo contained only hydrofluoroalkane propellant. The primary outcome was death or oxygen dependency at discharge from NICU. Secondary outcomes included death, severe BPD and neurodevelopmental outcomes at 18 months' PMA and three years of age. Complications of preterm birth (grade 3 or 4 IVH; periventricular leukomalacia (PVL); NEC; sepsis and ROP) were reported in combination with deaths.

Townsend 1998: The study by Townsend and co‐workers enrolled 32 infants with GA less than 28 weeks and birth weight up to 1100 grams who were ventilator dependent with RDS. Infants were randomised to receive flunisolide or placebo in the dose of 500 µg three times a day delivered via spacer chamber. Treatment was begun at 48 to 96 hours of age and continued for 28 days or until extubation. Outcomes included were: need for systemic dexamethasone, duration of ventilation, duration of hospitalisation, duration of supplemental oxygen and the incidence of adverse events (hyperglycaemia, hypertension, weight gain).

Yong 1999: The study by Yong and co‐workers enrolled 40 infants born at less than 32 weeks' GA and requiring mechanical ventilation at birth within 18 hours after birth. Infants were randomly assigned to receive fluticasone propionate or placebo. One puff (250 µg/puff) was administered twice a day for two weeks. There was no difference in the baseline characteristics between groups. Study outcomes included frequency of BPD at 28 days of life and at 36 weeks' PMA, duration of respiratory support, need for systemic corticosteroids, mortality, duration of hospitalisation, successful extubation by seven and 14 days of age, pulmonary function tests (compliance and resistance), inflammatory markers in the tracheal aspirates and incidence of adverse events.

Excluded studies

See 'Characteristics of excluded studies'.

Four trials were excluded (Kovacs 1998; Beresford 2002; Dugas 2005; Yeh 2016). Beresford 2002 was excluded as infants had to be receiving supplemental oxygen at 36 weeks' PMA at the time of randomisation. Denjean 1998 was excluded in previous versions of this review as both ventilated and non‐ventilated infants were included and we were unable to obtain data for ventilated infants from the authors. In a deviation from our protocol for this update we included all infants on any form of assisted ventilatory support (nasal CPAP/IMV or endotracheal IMV). Dugas 2005 was excluded as infants were randomised between 28 and 60 days of age (late). Kovacs 1998 was excluded as investigators evaluated the impact of a combination of systemic and inhaled corticosteroid for prevention of CLD. Yeh 2016 compared the effect of intratracheal administration of surfactant/budesonide with that of surfactant alone on the incidence of death or BPD.

Allocation

The risk of bias regarding the random sequence generation was low in Bassler 2015 and Nakamura 2016 and the risk was unclear in the remaining eight studies.

The allocation concealment was of low risk of bias in all studies except for Denjean 1998 in which no information was provided.

Blinding

The risk of performance bias was low in nine studies (no information provided in Denjean 1998) and the risk of detection bias was low in three studies and unclear in seven studies.

Incomplete outcome data

Attrition bias was low in nine studies and unclear in one study (Cole 1999).

Exclusions after randomisation:

In Bassler 2015 four infants in the budesonide group and three infants in the placebo group had unknown outcomes because of withdrawal of consent or right to use the data.

In Cole 1999, three infants were withdrawn before the study drug was administered (two due to sepsis and one due to prior receipt of systemic glucocorticoid therapy). In the treatment group, one infant had been withdrawn prior to 28 days of life and eight more were withdrawn by 36 weeks' PMA. Similarly, in the placebo group, two infants had been withdrawn by 28 days of age and four more were withdrawn by 36 weeks' PMA. Reasons for these later withdrawals were not described. The outcome data for withdrawn infants were not provided. Thus, outcomes were reported at 28 days for 122 treated and 128 control infants, and at 36 weeks' PMA for 114 treated and 124 control infants.

In Denjean 1998 178 infants were initially randomised, but informed consent was either not obtained or withdrawn for five infants, leaving 173 infants in the trial. During the study the randomisation code was broken in three cases because of severe deterioration.

One infant in the placebo group was withdrawn from Merz 1999 due to severe sepsis.

Selective reporting

In only one study was the protocol available to us (Bassler 2015). In the other studies the study protocol was not available to us and we could not judge if there were any deviations from the protocol or not.

Other potential sources of bias

We could not ascertain if there was other bias in one study for which only the abstract was available to us (Townsend 1998). For the remaining studies we could not identify any other potential sources of bias.

Early inhaled steroids (< 2 weeks) vs. placebo among all randomised (Comparison 1)

If only one study is included in an analysis tests for heterogeneity are not applicable.

Primary outcome

CLD at 36 weeks' PMA (Outcome 1.1):

Five trials enrolling 429 neonates reported on the incidence of CLD at 36 weeks' PMA among all randomised (Cole 1999; Fok 1999; Merz 1999; Yong 1999; Jangaard 2002). There was no significant difference in the incidence of CLD at 36 weeks' PMA in the inhaled steroid group versus the placebo group (typical RR 0.97, 95% CI 0.62 to 1.52; typical RD −0.00, 95% CI −0.07 to 0.06). There was no heterogeneity for this outcome for RR (I² = 11%) nor for RD (I² = 0%). Analysis 1.1Figure 4.

Figure 4

---

---

Forest plot of comparison: 1 Early inhaled steroids (< 2 weeks) vs. placebo (among all randomised), outcome: 1.1 CLD at 36 weeks' PMA.

Secondary outcomes

CLD at 28 days of age (Outcome 1.2):

Five trials enrolling 429 neonates reported on the incidence of CLD at 28 days of age among all randomised (Cole 1999; Fok 1999; Merz 1999; Yong 1999; Jangaard 2002). There was no significant difference in the incidence of CLD at 28 days (typical RR 1.08, 95% CI 0.85 to 1.38; typical RD 0.03, 95% CI −0.08 to 0.14). There was no heterogeneity for this outcome for RR (I² = 4%) nor for RD (I² = 23%). Analysis 1.2.

Death by 28 days of age (Outcome 1.3):

Five trials enrolling 429 neonates reported on the incidence of death by 28 days of age among all randomised (Cole 1999; Fok 1999; Merz 1999; Yong 1999; Jangaard 2002). There was no significant effect on death by 28 days of age (typical RR 0.66, 95% CI 0.39 to 1.14; typical RD −0.04, 95% CI −0.09 to 0.01). There was no heterogeneity for this outcome for RR (I² = 0%); however, there was moderate heterogeneity for RD (I² = 59%). Analysis 1.3.

Death by 36 weeks' PMA (Outcome 1.4):

Six trials enrolling 1285 neonates reported on the incidence of death by 36 weeks' PMA among all randomised (Cole 1999; Fok 1999; Merz 1999; Yong 1999; Jangaard 2002; Bassler 2015). No significant effect on mortality by 36 weeks' PMA was noted (typical RR 1.07, 95% CI 0.82 to 1.40; typical RD 0.01, 95% CI −0.03 to 0.05). There was moderate heterogeneity for this outcome for RR (I² = 51%); and there was low heterogeneity for RD (I² = 48%). Analysis 1.4.

Death by, or CLD at, 28 days of age (Outcome 1.5):

Five trials enrolling 429 neonates reported on the incidence of death by or CLD at 28 days of age among all randomised (Cole 1999; Fok 1999; Merz 1999; Yong 1999; Jangaard 2002). There was no significant difference between the groups for the combined outcome of death by, or CLD at, 28 days of age (typical RR 0.96, 95% CI 0.80 to 1.14; typical RD −0.02, 95% CI −0.11 to 0.07). There was no statistically significant heterogeneity for this outcome (RR I² = 0%; RD I² = 0%). Analysis 1.5.

Death by, or CLD at, 36 weeks' PMA (Outcome 1.6):

Six trials enrolling 1285 neonates reported on this outcome (Cole 1999; Fok 1999; Merz 1999; Yong 1999; Jangaard 2002; Bassler 2015). There was a significant difference noted for the combined outcome of death by, or CLD at, 36 weeks' PMA (typical RR 0.86, 95% CI 0.75 to 0.99; typical RD −0.06, 95% CI −0.11 to −0.00) (P = 0.04); NNTB 17 (95% CI 9 to infinity). There was no statistically significant heterogeneity for this outcome (RR I² = 0%; RD I² = 0%). Analysis 1.6; Figure 5.

Figure 5

---

---

Forest plot of comparison: 1 Early inhaled steroids (< 2 weeks) vs. placebo (among all randomised), outcome: 1.6 Death by or CLD at 36 weeks' PMA.

Survival to hospital discharge without CLD (Outcome 1.7):

One trial enrolling 86 infants reported on this outcome (Denjean 1998). There was no significant difference between the groups for survival to discharge without CLD (RR 1.50, 95% CI 0.83 to 2.72; RD 0.14, 95% CI −0.06 to 0.34). Analysis 1.7.

Death during hospital stay (Outcome 1.8):

One trial enrolling 86 infants reported on this outcome (Denjean 1998). There was no significant difference between the groups for death during hospital stay (RR 1.75, 95% CI 0.55 to 5.55; RD 0.07, 95% CI −0.07 to 0.21). Analysis 1.8.

Culture‐proven infection (Outcome 1.9):

Two trials reported on 896 neonates for the incidence of culture‐proven infection (positive blood or CSF culture) (Yong 1999; Bassler 2015). There was no statistically significant difference in the incidence of culture‐proven infection based on this definition (typical RR 1.17, 95% CI 0.97 to 1.41; typical RD 0.05, 95% CI −0.01 to 0.11). There was low heterogeneity for this outcome for RR (I² = 34 %) and for RD (I² = 47 %).

Four trials reported on 225 neonates for the incidence of culture‐proven infection (positive blood culture) (Denjean 1998; Fok 1999; Jonsson 2000; Jangaard 2002). There was no statistically significant difference in the incidence of culture‐proven infection based on this definition (typical RR 1.17, 95% CI 0.83 to 1.65; typical RD 0.05, 95% CI −0.06 to 0.16). There was no statistically significant heterogeneity for this outcome (RR I² = 0%; RD I² = 0%).

For the two outcomes (positive blood culture or CSF culture and positive blood culture) combined (N = 1121 neonates) there was no statistically significant difference for culture‐proven infection (typical RR 1.17, 95% CI 0.99 to 1.38; typical RD 0.05, 95% CI −0.00 to 0.11). There was no statistically significant heterogeneity for this outcome (RR I² = 0%; RD I² = 0%). Analysis 1.9.

Hyperglycaemia (Outcome 1.10):

Three trials enrolling 116 neonates reported on the incidence of hyperglycaemia (Fok 1999; Merz 1999; Yong 1999). There was no significant difference in the incidence of hyperglycaemia (typical RR 0.84, 95% CI 0.49 to 1.44; typical RD −0.05, 95% CI −0.19 to 0.10). There was no statistically significant heterogeneity for this outcome (RR I² = 0%; RD I² = 0%). Analysis 1.10.

Bassler 2015 found no significant difference in the occurrence of hyperglycaemia requiring insulin treatment between the two groups (n = 856) (RR 0.97, 95% CI 0.74 to 1.27; RD −0.01, 95% CI −0.06 to 0.05). Tests for heterogeneity were not applicable. Analysis 1.10.

Hypertension (Outcome 1.11):

Three trials enrolling 116 neonates reported on the incidence of hypertension (Fok 1999; Merz 1999; Yong 1999). There was no statistically significant difference in the incidence of hypertension between groups (typical RR 1.20, 95% CI 0.36 to 3.99; typical RD 0.01, 95% CI −0.09 to 0.12). Test for heterogeneity was not applicable for RR (two studies had no outcomes) and there was no heterogeneity for RD (I² = 0%). Analysis 1.11.

Bassler 2015 reported no significant difference in the occurrence of hypertension requiring treatment between the two groups (n = 856) (RR 0.58, 95% CI 0.21 to 1.57; RD −0.01, 95% CI −0.03 to 0.01). Tests for heterogeneity were not applicable. Analysis 1.11.

Gastrointestinal bleeding (Outcome 1.12):

One trial enrolling 253 neonates reported on the incidence of gastrointestinal bleeding (Cole 1999). There was no significant difference for this outcome between groups (RR 0.35, 95% CI 0.04 to 3.34; RD −0.01, 95% CI −0.05 to 0.02). Tests for heterogeneity were not applicable. Analysis 1.12.

Bassler 2015 did not report on gastrointestinal bleeding, but reported on gastrointestinal disorders which occurred in 30 cases (6.86%) in the inhaled budesonide group and in 32 cases (7.64%) in the placebo group. The contribution of gastrointestinal bleeding to the total number is not reported.

Cataract (Outcome 1.13):

One trial enrolling 253 neonates reported on the incidence of cataracts (Cole 1999). There was no significant difference in the incidence of cataracts between groups (RR 0.35, 95% CI 0.01 to 8.56; RD −0.01, 95% CI −0.03 to 0.01). Tests for heterogeneity were not applicable. Analysis 1.13.

Intraventricular haemorrhage (IVH) (Outcome 1.14):

Two trials enrolling 306 neonates reported on the incidence of IVH (Fok 1999; Cole 1999). There was no significant difference in the incidence of IVH between groups (typical RR 1.04, 95% CI 0.77 to 1.41; typical RD 0.02, 95% CI −0.09 to 0.12). There was no statistically significant heterogeneity for this outcome (RR I² = 0%; RD I² = 0%). Analysis 1.14.

Periventricular leukomalacia (PVL) (Outcome 1.15):

Two trials enrolling 306 neonates reported on the incidence of PVL (Fok 1999; Cole 1999). There was no significant difference in the incidence of PVL between groups (typical RR 1.43, 95% CI 0.59 to 3.46; typical RD 0.02, 95%CI −0.03 to 0.08). There was no statistically significant heterogeneity for this outcome (RR I² = 0%; RD I² = 0%). Analysis 1.15.

Brain injury (Outcome 1.16):

One trial reporting on 838 infants reported on the incidence of brain injury diagnosed based on the worst cranial ultrasound finding performed at or before 36 weeks' PMA (Bassler 2015). There was no significant difference in the incidence of brain injury between the two groups (RR 1.25, 95% CI 0.94 to 1.65; RD 0.04, 95% CI −0.01 to 0.10). Tests for heterogeneity were not applicable. Analysis 1.16

Necrotizing enterocolitis (NEC) (Outcome 1.17):

Three trials enrolling 1162 neonates reported on the incidence of NEC (Cole 1999; Fok 1999; Bassler 2015). There was no significant difference in the incidence of NEC between groups (typical RR 0.92, 95% CI 0.68 to 1.24; typical RD −0.01, 95% CI −0.05 to 0.03). There was low heterogeneity for this outcome for RR (I² = 31%) and for RD (I² = 40%). Analysis 1.17.

Retinopathy of prematurity (ROP) any stage (Outcome 1.18):

Two trials enrolling 306 neonates reported on the incidence of ROP (Fok 1999; Cole 1999). There was no significant difference in the incidence of ROP (typical RR 1.00, 95% CI 0.87 to 1.15; typical RD 0.00, 95% CI −0.09 to 0.09). There was no statistically significant heterogeneity for this outcome (RR I² = 0%; RD I² = 0%). Analysis 1.18

Three trials enrolling 1030 neonates reported on the incidence of ROP (Cole 1999; Fok 1999; Bassler 2015). There was no significant difference in the incidence of ROP (typical RR 1.06, 95% CI 0.93 to 1.21; typical RD 0.03, 95% CI −0.03 to 0.08). There was no statistically significant heterogeneity for this outcome (RR I² = 0%; RD I² = 0%). Analysis 1.18.

Patent ductus arteriosus (Outcome 1.19):

One trial enrolling 53 neonates reported on the incidence of PDA (Fok 1999). There was no significant difference in the incidence of PDA between the two groups (RR 0.82, 95% CI 0.57 to 1.17; RD −0.14, 95% CI −0.38 to 0.10). Tests for heterogeneity were not applicable. Bassler 2015 reported on PDA treated with drugs (reported un stratified RR 0.88, 95% CI 0.76 to 1.01) and with surgical ligation (reported un stratified RR 0.55, 95% CI 0.36 to 0.84). There was overlap between these groups and we are awaiting clarification by the first author prior to inclusion in the analysis. Analysis 1.19.

Reintubation (Outcome 1.20):

One trial reporting on 856 infants reported a significantly lower risk of reintubation in the steroid group compared with the control group (RR 0.58, 95% CI 0.35 to 0.96; RD −0.04, 95% CI −0.07 to −0.00; NNTB 25, 95% CI 14 to ∞) (Bassler 2015). Tests for heterogeneity were not applicable. Analysis 1.20.

Requirement for systemic steroids (Outcome 1.21):

Eight trials enrolling 1403 infants reported on the requirements for systemic steroids (Denjean 1998; Townsend 1998; Cole 1999; Fok 1999; Merz 1999; Yong 1999; Jangaard 2002; Bassler 2015). The need for systemic steroids was not statistically significantly different between groups (typical RR 0.89, 95% CI 0.77 to 1.02; typical RD −0.04, 95% CI −0.09 to 0.01). There was low heterogeneity for this outcome for RR (I² = 33%) and moderate heterogeneity for RD (I² = 63%). Analysis 1.21.

Failure to extubate within 14 days (Outcome 1.22):

Five trials enrolling 193 neonates reported on this outcome (Fok 1999; Merz 1999; Yong 1999; Jonsson 2000; Jangaard 2002). There was no significant difference noted for this outcome (typical RR 0.97, 95% CI 0.76 to 1.24; typical RD −0.02, 95% CI −0.15 to 0.12). There was moderate heterogeneity for this outcome for RR (I² = 61%) and high heterogeneity for RD (I² = 80%). Analysis 1.22.

The following outcomes (Analysis 1.23 to Analysis 1.32) all pertain to the trial by Nakamura and co‐workers (Nakamura 2016), who for all outcomes reported a combination of deaths and complications of preterm birth. As there is only one trial included in these analyses tests for heterogeneity are not applicable. Some denominators exclude infants for which the outcome could not be ascertained.

Death or oxygen dependency at discharge (Outcome 1.23):

This outcome was reported for all 211 randomised infants (Nakamura 2016). There was no significant difference between the two groups (RR 0.63, 95% CI 0.35 to 1.15; RD −0.08, 95% CI −0.18 to 0.02). Analysis 1.23.

Death or severe BPD (Outcome 1.24):

This outcome was reported for all 211 randomised infants (Nakamura 2016). There was no significant difference between the two groups (RR 0.93, 95% CI 0.70 to 1.25; RD −0.03, 95% CI −0.17 to 0.10). Analysis 1.24.

Death or grade 3 or 4 IVH (Outcome 1.25):

This outcome was reported for all 211 randomised infants (Nakamura 2016). There was no significant difference between the two groups (RR 1.05, 95% CI 0.65 to 1.68; RD 0.01, 95% CI −0.10 to 0.13). Analysis 1.25.

Death or PVL (Outcome 1.26):

This outcome was reported for all 211 randomised infants (Nakamura 2016). There was no significant difference between the two groups (RR 0.89, 95% CI 0.41 to 1.93; RD −0.01, 95% CI −0.10 to 0.07). Analysis 1.26.

Death or NEC (Outcome 1.27):

This outcome was reported for all 211 randomised infants (Nakamura 2016). There was no significant difference between the two groups (RR 0.97, 95% CI 0.46 to 2.06; RD −0.00, 95% CI −0.09 to 0.08). Analysis 1.27.

Death or sepsis (Outcome 1.28):

This outcome was reported for all 211 randomised infants (Nakamura 2016). There was no significant difference between the two groups (RR 0.79, 95% CI 0.44 to 1.40; RD −0.04, 95% CI −0.15 to 0.06). Analysis 1.28.

Death or ROP (stage not stated) (Outcome 1.29):

This outcome was reported for all 211 randomised infants (Nakamura 2016). There was no significant difference between the two groups (RR 1.05, 95% CI 0.79 to 1.40; RD 0.02, 95% CI −0.11 to 0.16). Analysis 1.29.

Death or neurodevelopmental impairment at 18 months PMA (Outcome 1.30):

This outcome was reported in 187 infants available at follow‐up (Nakamura 2016). There was no significant difference between the two groups (RR 1.09, 95% CI 0.70 to 1.70; RD 0.03, 95% CI −0.10 to 0.16). Analysis 1.30

Death or neurodevelopmental impairment at 3 years of age (Outcome 1.31):

This outcome was reported in 179 infants available at follow‐up (Nakamura 2016). There was no significant difference between the two groups (RR 1.03, 95% CI 0.68 to 1.56; RD 0.01, 95% CI −0.13 to 0.15). Analysis 1.31.

Death or cerebral palsy at 3 years of age (Outcome 1.32):

This outcome was reported in 190 infants available at follow‐up (Nakamura 2016). There was no significant difference between the two groups (RR 1.12, 95% CI 0.64 to 1.96; RD 0.02, 95% CI −0.09 to 0.14). Analysis 1.32.

Early inhaled steroids (< 2 weeks) vs. placebo among survivors (Comparison 2)

Primary outcome

CLD at 36 weeks' PMA (Outcome 2.1):

Six trials enrolling 1088 neonates reported on this outcome among survivors (Cole 1999; Fok 1999; Merz 1999; Yong 1999; Jangaard 2002; Bassler 2015). There was a significant difference in the incidence of CLD at 36 weeks' PMA among survivors (typical RR 0.76, 95% CI 0.63 to 0.93; typical RD −0.07, 95% CI −0.13 to −0.02; NNTB 14, 95% CI 8 to 50). There was no statistically significant heterogeneity for this outcome (RR I² = 0%; RD I² = 0%). Analysis 2.1; Figure 6.

Figure 6

---

---

Forest plot of comparison: 2 Early inhaled steroid (< 2 weeks) vs. placebo (among survivors), outcome: 2.1 CLD at 36 weeks' PMA.