Safety of Less Invasive Surfactant Administration in Preterm Infants: A Meta‐Analysis

ABSTRACT

Background

Less invasive surfactant administration (LISA) is increasingly used in preterm infants with respiratory distress syndrome (RDS). Understanding its safety profile compared to the INtubation‐SURfactant‐Extubation (INSURE) or endotracheal tube (ETT) techniques is essential.

Objective

To evaluate the safety profile of LISA compared to INSURE/ETT in preterm infants with RDS using evidence from randomized and non‐randomized studies.

Methods

A systematic literature review identified studies in MEDLINE, EMBASE, and CENTRAL (2010–2024). Arm‐level pooled incidence estimates were generated across RCTs and non‐randomized studies, and pairwise comparisons of LISA versus INSURE/ETT were conducted in randomized controlled trials (RCTs). Analyses utilized Generalized Linear Mixed Models with random‐effects. Subgroups included gestational age (≤ 34 weeks), surfactant type, antenatal steroid exposure (≥ 50%) and sedation and premedication use.

Results

Nineteen RCTs (2460 infants) and 26 non‐randomized studies (2360 infants) met inclusion criteria. Pairwise comparisons revealed no statistically significant differences between LISA and INSURE/ETT for apnea, bradycardia, cough, hypoxemia, or need for ≥ 2 attempts for successful catheter/ETT placement attempts; however, surfactant regurgitation/reflux occurred more frequently with LISA (OR = 2.29, 95% CI = 1.21–4.33). Adverse event rates were generally higher in non‑randomized studies, with hypoxemia the most frequently reported event across study types (≥ 24%). Subgroup findings were consistent with overall results.

Conclusion

To our knowledge, this is the first meta‐analysis evaluating LISA safety outcomes across both RCTs and non‐randomized studies. LISA demonstrates a safety profile broadly comparable to INSURE/ETT for most outcomes. Differences observed in observational studies likely reflect variability in reporting and clinical practice. Standardized training and procedure optimization may further enhance safety.

1. Introduction

Respiratory distress syndrome (RDS) remains a leading cause of morbidity and mortality in preterm neonates, primarily due to surfactant deficiency and immature pulmonary development [1]. Surfactant replacement therapy is a cornerstone of RDS management, traditionally administered via endotracheal intubation and mechanical ventilation [2]. While effective, this approach carries risks of ventilator‐induced lung injury and other iatrogenic complications [3]. In recent years, Less Invasive Surfactant Administration (LISA) has emerged as a promising alternative, enabling surfactant delivery via a thin catheter while maintaining spontaneous breathing on continuous positive airway pressure (CPAP). This technique has been endorsed by the European Consensus Guidelines as the preferred method for surfactant administration in spontaneously breathing preterm infants [4]. Meta‐analyses have demonstrated the efficacy of LISA in reducing bronchopulmonary dysplasia (BPD), mortality, and the need for mechanical ventilation compared to conventional methods such as INtubation‐SURfactant‐Extubation (INSURE) and endotracheal tube (ETT) administration [5, 6, 7].

While efficacy is well documented, the safety profile of LISA remains incompletely characterized. Procedural adverse events such as apnea, bradycardia, hypoxemia, surfactant regurgitation, and the need for multiple catheterization attempts have been reported inconsistently across studies. Existing meta‐analyses have either treated safety outcomes as secondary endpoints [5] or incorporated them within broader comparative frameworks, often without dedicated analysis [8]. The limited scope and heterogeneity of safety reporting in both randomized controlled trials and observational studies have hindered a comprehensive understanding of LISA's risk profile.

This systematic review and meta‐analysis aim to synthesize safety data from RCTs and non‐randomized studies evaluating LISA in preterm infants with or at risk of RDS. By incorporating subgroup analyses and comparing findings across study designs, we seek to elucidate the procedural risks associated with LISA and inform best practices for its safe implementation in neonatal care. The review protocol was registered in PROSPERO (CRD42024599816).

2. Methods

2.1. Eligibility Criteria and Study Design

We included randomized controlled trials (RCTs) and non‐randomized studies that evaluated LISA in preterm infants (< 37 weeks gestational age) with established or suspected RDS. Studies were eligible if they reported at least one predefined safety outcome during or immediately after the surfactant administration procedure. The exact assessment window was extracted as reported in each study, and no uniform post‐procedure time window was imposed retrospectively because timing definitions varied across studies. Studies enrolling term infants, those with congenital anomalies affecting respiratory function, or lacking safety outcome data were excluded. The intervention of interest was LISA, defined as surfactant delivery via a thin catheter while the infant remains on non‐invasive respiratory support (e.g., CPAP). For RCTs, comparators of interest included surfactant administration INSURE or ETT techniques. Studies using alternative terminology (e.g., MIST, SurE, MISA, NISA, TCA) were eligible in the SLR if the intervention met the core definition.

2.2. Safety Outcomes

The safety outcomes of interest were procedural adverse events occurring during or immediately after surfactant administration, including: apnea (cessation of breathing requiring intervention); cough, sneezing, choking, vomiting; hypoxemia/desaturation (SpO₂ < 80% or as defined by study authors); need for ≥ 2 attempts for successful catheter/ETT placement; need for intubation during the procedure; need for PPV during the procedure; surfactant regurgitation/reflux. Bradycardia was analyzed in two ways: author‐defined bradycardia and bradycardia defined as heart rate < 100 bpm. The threshold‐based definition was considered the more standardized outcome, whereas author‐defined bradycardia was retained to preserve data from studies that reported bradycardia without specifying a uniform heart‐rate threshold. The two definitions were analyzed separately and were not combined into a single pooled estimate.

2.3. Literature Search and Study Selection

A comprehensive search was conducted in MEDLINE, EMBASE, and CENTRAL using controlled vocabulary and keywords related to preterm infants, surfactant therapy, and LISA. For randomized controlled trials, the search was limited to studies published in English from 2020 to September 10, 2024, supplementing a previously published Cochrane review [5]; for non‑randomized and observational studies, the search spanned database inception to October 4, 2024. Full search strategies are provided in Supporting Information S1. Study selection was conducted in two stages using DistillerSR®, an online systematic review platform that facilitates AI‐assisted screening and reviewer collaboration. In the first stage, duplicate records were removed, and titles and abstracts were screened by a primary reviewer against predefined eligibility criteria. Any records excluded at this stage but flagged by the AI algorithm with a high probability of relevance (> 85%) were reassessed by a second reviewer. In the second stage, full‐text articles were independently reviewed by two researchers, with any disagreements resolved through discussion or adjudication by a third reviewer. The overall selection process was documented using a PRISMA flow diagram.

2.4. Data Extraction and Risk of Bias

Data was extracted using a standardized template by two independent reviewers. Extracted variables included study design, sample size, gestational age, intervention details, comparator, and safety outcomes. For RCTs, risk of bias was assessed using Revised Cochrane Risk of Bias Tool (RoB 2) [9]. Non‐randomized studies were evaluated using the Newcastle‐Ottawa scale (NOS) [10]. Details of the tools are provided in Supporting Information S2.

2.5. Meta‐Analysis Feasibility Assessment

All studies identified through the systematic literature review were evaluated for inclusion in the meta‐analysis based on the comparability of study design, patient populations, intervention characteristics, and outcome definitions. Particular attention was given to consistency in reporting safety endpoints across studies. Clinical experts were consulted to assess the appropriateness of pooling data and to ensure that methodological decisions reflect clinical relevance.

2.6. Statistical Analysis and Reporting Bias

Treatment effects for dichotomous variables were expressed as odds ratios (ORs) with 95% confidence intervals (CIs). Pairwise meta‐analyses comparing LISA with INSURE or ETT were conducted using generalized linear mixed models (GLMMs). Analyses were performed in R (version 4.4.3) using the metabin function from the meta package. Random‐effects models were applied as the primary approach to account for expected heterogeneity across studies. Arm‐level pooled analyses were conducted for both randomized and non‐randomized studies to estimate the incidence of individual safety events, using the mataprop package in R. Statistical heterogeneity was assessed using the Q statistic, I², τ², and prediction intervals, with I² thresholds defined as: 0%–24% (none), 25%–49% (low), 50%–74% (moderate), and 75%–100% (high). Funnel plots and Harbord's test were generated for outcomes with ≥ 10 studies to evaluate potential publication bias and small‐study effects. Outcomes based on fewer than 10 studies were considered exploratory.

2.7. Subgroup Analyses

To explore potential modifiers of safety outcomes, subgroup analyses were conducted for outcomes reported in at least three studies. Predefined scenarios included: gestational age ≤ 34 weeks, use of Curosurf® as the surfactant product, no mandatory pre‐procedural medications, no sedation and at least 50% of the patients were exposed to antenatal steroids.

3. Results

3.1. Study Selection and Characteristics

The systematic literature review identified 814 records for RCTs and 1521 records for non‐ randomized studies through database searches. After duplicating removal and screening, 89 studies met the eligibility criteria for inclusion in the review, reporting on the safety or effectiveness of the LISA technique: 28 RCTs (Figure 1) and 61 non‐randomized studies (Figure 2). Following a feasibility assessment to confirm comparability of study design, patient populations, and outcome definitions, 19 RCTs and 26 non‐RCTs were included in the meta‐analysis for safety outcomes. Studies were excluded from quantitative synthesis if they did not report predefined safety outcomes or lacked sufficient data for pooling. Additional exclusions occurred when comparators were outside the scope (e.g., PAP continuation or alternative catheter types).

FIGURE 1.

PRISMA flow diagram for studies searches for RCTs. [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 2.

PRISMA flow diagram for studies searches for non‐RCTs and observational studies. [Color figure can be viewed at wileyonlinelibrary.com]

As shown in (Table 1), among the 19 RCTs enrollment periods ranged from 2007 to 2024, with sample sizes between 31 and 350 participants. Fourteen RCTs enrolled infants with gestational age (GA) < 34 weeks, and five extended inclusion up to 36–37 weeks. Seven trials included extremely preterm infants (< 28 weeks). Timing of surfactant administration was reported in 13 RCTs, with initiation within 6–12 h of birth in nine studies and studies within 2 h in four studies. FiO₂ thresholds were specified in 11 trials, typically ≥ 30%–40%. CPAP or NCPAP was used prior to LISA in all RCTs, and delivery‐room intubation was prohibited in 15 trials.

TABLE 1.

Characteristics and main eligibility criteria of RCTs and non‐randomized trials included in meta‐analysis.

Author year study name, trial ID†	Country	Enrollment years	Treatment groups and sample size	Gestational age	RDS status	Age at enrollment	FiO2 requirement	CPAP	Intubation in the delivery room
Randomized controlled trials
Akcay 2021 [11]	Turkey	NR	LISA: 42INSURE: 36Total: 78	< 34 weeks	Infants with signs of RDS	Within 2 h of life	40% if GA > 26 weeks> 30% if GA < 26 weeks	NCPAP	NR
Anand 2022 [12](CTRI/2020/05/025360)	India	NR	LISA: 74INSURE: 76Total: 150	26–34 weeks	Infants with signs of RDS	Within 6 h of life	> 30%	NCPAP	Not allowed
Auer–Hackenberg 2022 [13](NCT05024435)	Austria	2019–2020	LISA: nasogastric tube: 16LISA: LISAcath® tube: 15Total: 31	< 37 weeks	NR	NR	NR	NR	NR
Bao 2015 [14] (ChiCTR–ICR–15006001)	China	2012	LISA: 47INSURE: 43Total: 90	28–32 weeks	Infants with RDS and in need of surfactant administration within 2 h after birth	NR	NR	CPAP	Not allowed
Choupani 2018 [15]	Iran	2016–2017	LISA: 52INSURE: 52Total: 104	28–37 weeks	Infants with signs of RDS	Within 1 h of life	NR	NCPAP	Not allowed
Dargaville 2021 [16] (OPTIMIST–A, ACTRN1261100091694)	Multiple countries: Australia, Canada, Israel, New Zealand, Qatar, Singapore, Slovenia, the Netherlands, Turkey, UK, US	2011–2020	LISA: 241CPAP: 244a	25–28 weeks + 6 days	Infants with RDS	Within 6 h of life	> 30% within the first 6 h of life	CPAP or NIPPV	NR
Göpel 2011 [17] (The AMV trial, ISRCTN05025922)	Germany	2007–2010	LISA: 108CPAP: 112a	26–28 weeks + 6 days	Infants were enrolled irrespective of their respiratory status	Within 12 h of life	NR	CPAP	Allowed
Halim 2019 [18]	Pakistan	2017	LISA: 50INSURE: 50Total: 100	< 34 weeks	Infants with RDS	Within 12 h of life	> 40%	NR	Not allowed
Han 2020 [19] (MISA study, NCT04077333)	China	2017–2018	LISA: 176INSURE: 168Total: 344	< 31 weeks + 6 days	Infants with RDS or signs of RDS and in need of surfactant administration within 6 h after birth	NR	> 40%	NCPAP	Not allowed
Jena 2019 [20]	India	2013–2017	LISA: 175INSURE: 175Total: 350	≤ 34 weeks	Infants with RDS	Within 6 h of life	> 40%	CPAP	Not allowed
Kanmaz 2013 [21](NCT01329432)	Turkey	2010–2011	LISA: 100INSURE: 100Total: 200	< 32 weeks	Infants with RDS	NR	NR	NR	Not allowed
Katheria 2023 [22] (CaLI, NCT04209946)	US	2020–2022	LISA: 92CPAP: 88a	24–29 weeks + 6 days	NR	Within 1 h of life	NR	CPAP	Not allowed
Kribs 2015 [23] (NINSAPP, ISRCTN64011614/NCT00751959)	Germany	2009–2012	LISA: 107ETT: 104Total: 211	23–26 weeks + 6 days	Infants with RDS	Within 10 min to 2 h of life	> 30%	NR	NR
Mirnia 2013 [24]	Iran	2010–2012	LISA: 66INSURE: 70Total: 136	27–32 weeks	Infants with RDS	NR	NR	NR	NR
Mishra 2023 [25] (CTRI/2022/01/039147)	India	NR	LISA: 75INSURE: 75Total: 150	28–36 weeks	Infants with RDS	NR	NR	NR	Not allowed
Mosayebi 2017 [26] (IRCT2014080716937N4)	Iran	2013–2014	LISA: 27INSURE: 26Total: 53	28–36 weeks + 6 days	Infants with RDs	Within 24 h of life	NR	NR	Not allowed
Olivier 2017 [27]	Canada	2014–2016	LISA: 24CPAP: 21a	32–36 weeks + 6 days	Infants with RDS	Within 24 h of life	≥ 35%	NCPAP	Not allowed
Sabzehei 2022 [28](IRCT20160523028008N81)	Iran	2019–2020	LISA: 56INSURE: 56Total: 112	28–36 weeks	Infants with RDS	NR	NR	NR	Not allowed
Yang 2020 [29]	China	2017–2018	LISA: 47INSURE: 50Total: 97	33–36 weeks + 6 days	Infants with RDS	Within 12 h of life	> 40%	NCPAP	Not allowed
Non–randomized studies
Alsina–Casanova [30] 2024	Spain	2017–2022	LISA: 174ETT: 115a	23–33 weeks + 6 days	Infants with RDS	NR	NR	NR	NR
Bensouda 2022 [31]	Canada	2017–2021	LISA: 101	< 37 weeks	Infants with RDS	Within 72 h of life	≥ 30%	CPAP or NIV	Not allowed
Bhattacharya 2019 [32]	Canada	2016–2018	LISA: 43	≥ 28 weeks	Infants with RDS	NR	NR	NR	NR
Bhattacharya 2023 [33]	Canada	2016–2020	LISA via Angiocath: 46LISA via MAC: 93Total: 139	≥ 26 weeks	Infants with RDS	NR	NR	NR	Not allowed
Dargaville 2011 [34]	Australia	2009–NR	LISA: 25CPAP: 173a	29–34 weeks	Infants with RDS	Within 24 h of life	≥ 35%	CPAP	Not allowed
Dargaville 2013 [35]	Australia	2009–2011	LISA: 61CPAP: 97a	25–32 weeks	Infants with RDS	NR	≥ 30% if GA 25–28 weeks ≥ 35% if GA 29–32 weeks	CPAP	Not allowed
Dargaville 2018 [35]	Australia	2004–2013	LISA: 37	29–32 weeks + 6 days	Infants with RDS	Within 24 h of life	≥ 35%	CPAP	Not allowed
de Kort 2020 [36]	Netherlands	2016–2018	LISA: 86	< 32 weeks	Infants with RDS	NR	NR	NR	NR
Elbaz 2022 [37]	Israel	2014–2017	LISA: 45ETT: 62a	23–36 weeks + 6 days	Infants with RDS	NR	NR	NCPAP	NR
Gunes 2022 [38]	Turkey	2014–2015	LISA: 59INSURE: 55a	NR	Infants with RDS	NR	NR	NR	Not allowed
Harthy 2023 [39]	Canada	2021–2022	MIST–BLEScath™: 15MIST–SurfCath™: 22Total: 37	≥ 26 weeks	Infants with RDS	NR	≥ 30%	NR	NR
Klebermass–Schrehof 2013 [40]	Austria	2009–2011	LISA: 224CPAP: 182a	23–27 weeks + 6 days	Infants with RDS	NR	NR	NR	Not allowed
Klein 2024 [41]	Germany	2018	LISA: 57	22–27 weeks + 6 days	NR	NR	NR	NR	NR
Konda 2022 [42]	India	2020	LISA Orogastric tube: 25LISA Endotracheal tube (un–cuffed): 25Total: 50	28–34 weeks	Infants with RDS	Within 6 h of life	NR	NR	Not allowed
Maiwald 2021 [43]	Germany	2019	LISA: 20	NR	Infants with RDS	NR	≥ 30%	NR	NR
Mendoza 2021 [44]	Australia	2016–2020	LISA: 23	≥ 32 weeks	Infants with RDS	Within 24 h of life	≥ 35%	CPAP	NR
Patrikar 2023 [45]	India	2020–2021	LISA: 104	< 35 weeks	Infants with RDS	NR	30%	NCPAP	Not allowed
Pichler 2023 [46] (DRKS00022317)	Austria, Germany	2020–2021	LISA: 82	22–31 weeks + 6 days	Infants with signs of RDS	NR	NR	NR	Not allowed
Permall 2024 [47] (NCT03989960)	China	2018–2019	LISA + SNIPPV: 32InSURE + CPAP: 57a	< 36 weeks + 6 days	Infants with early symptoms of RDS or infants clinically diagnosed as RDS	NR	NR	NR	Not allowed
Ramos–Navarro 2016 [48] (NCT02611284)	Spain	2013 and 2014: LISA2012 to 2013: INSURE	LISA: 30INSURE: 30a	< 32 weeks	NR	Within 72 h of life	NR	NCPAP	Not allowed
Roberts 2021 [49]	Australia	2018	LISA: 122	23–40 weeks	Infants with RDS	NR	30% to 35% or higher	CPAP	Not allowed
Shetty 2024 [50]	UK	2018–2022	LISA–Delivery suite: 54LISA–NICU: 26Total: 80	< 32 weeks	NR	NR	NR	NR	NR
Sk 2023 [51]	India	2019–2022	LISA: 110	28–33 weeks + 6 days	Infants with RDS	NR	NR	NR	Not allowed
Szczapa 2020 [52]	Poland	2018–2019	LISA: 500	NR	NR	NR	NR	NR	NR
Thodika 2022 [53]	UK	2014–2019: Historical control 2019–2021: Case series	LISA: 99 ETT (if needed): 99a	26 to 34 weeks + 6 days	NR	NR	NR	NR	Not allowed
Xu 2022 [54]	China	2018–2019	LISA: 46 INSURE: 102a	< 32 weeks	Infants with RDS	Within 2 h of life	≥ 30%	NCPAP	Not allowed

Note: † Study name and register ID if reported.

Abbreviations: AMV, The Avoiding Mechanical Ventilation; CPAP, Continuous Positive Airway Pressure; ETT, Endotracheal Tube; INSURE, INtubation‐SURfactant‐Extubation; LISA, Less Invasive Surfactant Administration; MAC, Multi Access catheter®; MIST, Minimally Invasive Surfactant Therapy; NICU, Neonatal Intensive Care Unit; NR, Not reported; RDS, Respiratory distress syndrome; SNIPPV, Synchronized Nasal Intermittent Positive Pressure Ventilation; UK, United Kingdom; US, United States.

^a Treatment group was not included for meta‐analysis.

Across the 26 non‐randomized studies, sample sizes ranged from 20 to 500 infants (median = 106). 21 studies focused on GA < 32 weeks, while five included infants up to 36 weeks. FiO₂ criteria were reported in 14 studies, generally ≥ 30%–35%. CPAP was the predominant mode of support, and intubation at birth was restricted in 22 studies. Supporting Information S3 provides additional information on study characteristics and quality assessments. Supporting Information S1: Table 1 summarizes baseline characteristics, with gestational ages ranging from 23 to 37 weeks and birth weights between approximately 1000 g and 2200 g. Antenatal steroid exposure varied considerably across studies, from 3% to over 90%.

Treatment and procedural characteristics varied across studies (Supporting Information S1: Tables 2 and 3). Among RCTs, catheter type was reported for 25 of 26 LISA groups and included feeding, gastric, nasogastric, or gavage tubes in 15 groups; vascular or Angiocath catheters in four groups; semi‐rigid or purpose‐built LISA catheters in two groups; and umbilical catheters in one group. In non‐randomized studies, catheter type was reported for 27 of 30 LISA groups and included Angiocath or vascular catheters, feeding or gastric tubes, multi‐access catheters, LISAcath®, Surfcath^TM, BLEScat^TM, Neofact®, and other catheter types. Forceps use also varied, with Magill or ophthalmic forceps reported in nine RCT LISA groups and seven non‐randomized LISA groups, while several studies explicitly reported no forceps use. Surfactant products and pre‐procedural practices also differed, including variation in caffeine use, atropine, sucrose, fentanyl, morphine, and whether sedation or analgesia was allowed.

Supporting Information S1: Table 4 details sample sizes and the incidence of predefined safety outcomes, including apnea, bradycardia (author‐defined and HR < 100 bpm), hypoxemia, cough or gagging, multiple catheter attempts, surfactant regurgitation, and procedural interventions such as PPV or intubation. Quality assessments are summarized in Supporting Information S1: Tables 5–7. For RCTs, six studies were rated at low risk, 10 had some concerns, and three were high risk, mainly due to missing outcome data or deviations from intended interventions. Non‐randomized studies scored 5–7 out of 9 for comparative designs and 4–6 out of 6 for non‐comparative designs on the NOS, indicating overall moderate to high methodological quality.

3.2. Arm Level Pool Analysis

Arm‐level pooled analyses demonstrated that the incidence of procedural adverse events varied substantially across study designs (Table 2). In RCTs (19 studies; 2460 infants), apnea and bradycardia occurred in fewer than 10% of infants receiving LISA (incidence [95% CI]: apnea 2.66% [0.3, 20.06], bradycardia 7.0% [3.48, 13.59], bradycardia defined as HR < 100 bpm 8.27% [3.4, 18.73]). Cough or gagging, surfactant regurgitation, need for positive pressure ventilation and multiple catheterization attempts occurred in 10%–20% of infants. Hypoxemia was the most frequent event, affecting approximately one‐third of infants (34.04% [14.87, 60.4]). Among non‐randomized studies (26 studies; 2360 infants), pooled incidence estimates for LISA were generally higher than those observed in RCTs. The largest differences were observed for apnea (15.95% [8.51, 27.9] vs. 2.66% [0.3, 20.06] in RCTs) and need for ≥ 2 attempts (24.51% [18.59, 31.6] vs. 17.35% [11.85, 24.69] in RCTs). Estimates for hypoxemia (34.77% [23.89, 47.52]) and regurgitation (18.75% [11.46, 29.16]) were comparable to RCT estimates.

TABLE 2.

Arm‐level pooled incidence of safety outcomes using random‐effects model in RCTs and non‐randomized studies.

Outcome	RCT LISA	Non‐randomized LISA
	Incidence (%) [95% confidence interval]	Incidence (%) [95% confidence interval]
Apnea	2.66 [0.3, 20.06]	15.95 [8.51, 27.9]
Bradycardia (author definition)	7 [3.48, 13.59]	10.13 [6.64, 15.16]
Bradycardia (heart rate < 100 bpm)	8.27 [3.4, 18.73]	12.04 [7.84, 18.06]
Cough, sneezing, choking, vomiting	17.51 [7.12, 37]	19.16 [13.88, 25.84]
Hypoxemia/oxygen desaturation	34.04 [14.87, 60.4]	34.77 [23.89, 47.52]
Need for ≥ 2 attempts for successful catheter/ETT placement	17.35 [11.85, 24.69]	24.51 [18.59, 31.6]
Need for intubation during the procedure	—	2.72 [1.37, 5.35]
Need for PPV during the procedure	10.57 [7.03, 15.59]	12.55 [5.14, 27.53]
Surfactant regurgitation/reflux	15.94 [7.04, 32.17]	18.75 [11.46, 29.16]

Note: Results of the arm‐level pooled analysis are presented as incidence and 95% confidence interval, where incidence reflects the proportion of patients experiencing the safety event. * For studies reporting multiple groups for LISA (e.g., different catheters), all eligible groups were included for analysis.

Abbreviations: IVH, Intraventricular Hemorrhage; LISA, Less Invasive Surfactant Administration; PPV, Positive Pressure Ventilation; RCT, Randomized Controlled Trial.

Heterogeneity patterns differed substantially by study design and outcome. Hypoxemia showed the highest heterogeneity for LISA (I² = 95% in RCTs and 93% in non‐randomized studies), indicating non‐ uniform rates across trials. Apnea was consistent in RCTs (I² = 0%) but highly variable in non‐randomized cohorts (I² = 91%). Bradycardia remained highly heterogeneous in all settings (I² = 71%–93%) (Supporting Information S3, Supporting Information S1: Tables 8, 9).

3.3. Pairwise Comparison

Seven safety outcomes comparing LISA with INSURE/ETT in RCTs were analyzed (Figure 3). 12 trials, including 1747 infants, contributed data to these analyses. The meta‐analysis showed that the overall safety profile of LISA was comparable to INSURE/ETT for most outcomes. No statistically significant differences were observed for most safety outcomes. These included apnea (OR: 1.35 [0.63–2.88]), bradycardia by author definition (OR: 1.04 [0.53–2.04]), bradycardia defined as heart rate < 100 bpm (OR: 1.17 [0.51–2.69]), cough or gagging (OR: 2.31 [0.55–9.75]), hypoxemia or oxygen desaturation (OR: 0.85 [0.25–2.85]), and need for ≥ 2 attempts for successful catheter or ETT placement (OR: 1.17 [0.77–1.79]). Surfactant regurgitation/reflux was the only outcome that occurred significantly more frequently with LISA (OR: 2.29 [1.21–4.33]).

FIGURE 3.

Forest plots for pairwise comparison of safety outcomes between LISA and INSURE in randomized controlled trials. *Forest plots display ORs with 95% CIs for each outcome. The vertical dashed line represents the line of no effect (OR = 1). Estimates to the left of the line favor LISA over INSURE. All analyses were conducted using both common and random‐effects models. (A) Apnea. (B) Bradycardia by Author Definition. (C) Bradycardia (heart rate < 100 bpm). (D) Cough, sneezing, choking, vomiting. (E) Hypoxemia/oxygen desaturation. (F) Need for ≥ 2 attempts for successful catheter/ETT placement. (G) Surfactant regurgitation/reflux. [Color figure can be viewed at wileyonlinelibrary.com]

Heterogeneity varied across outcomes: it was high for hypoxemia (I² = 91.4%), moderate for bradycardia and surfactant regurgitation (I² ≈ 53–59%), and negligible for the remaining outcomes (I² = 0%). Because all outcomes were reported in fewer than 10 trials, assessment of publication bias using funnel plots was considered exploratory (Supporting Information S3, Supporting Information S1: Figures 1–7).

3.4. Subgroup Analysis

Subgroup analyses were conducted in both arm‐level pooled incidence analysis and pairwise comparison analysis. Results for pairwise comparisons between LISA and INSURE/ETT in select subgroups were largely consistent with the overall population, with surfactant regurgitation as the only outcome significantly higher for LISA (Supporting Information S1: Table 8). This effect was amplified in infants ≤ 34 weeks (OR 3.41 [1.73–6.74]) and in studies without premedication or sedation (OR 3.59 [2.02–6.38]), while other outcomes such as apnea and hypoxemia showed no consistent subgroup effect. Curosurf subgroup appeared to have a comparable or lower safety risk than the overall population. Arm‐level pooled incidence in RCTs confirmed hypoxemia as the most frequent event for LISA, exceeding 30% across scenarios and rising further in infants ≤ 34 weeks (44.15% [17.62, 74.50]) (Supporting Information S1: Table 9). Regurgitation varied with procedural factors, decreasing without sedation (8.17% [1.54, 33.63] vs 15.94 [7.04, 32.17] in overall population). Apnea remained infrequent in subgroups analysis with a higher incidence observed in Curosurf subgroup (7.98% [2.86, 20.37]). Non‐randomized studies of LISA reported higher rates for most outcomes, except for hypoxemia, where most subgroups showed decreasing incidence, particularly in the Curosurf subgroup (25.03% [12.48, 43.87]) (Supporting Information S1: Table 10).

4. Discussion

This meta‐analysis indicates that LISA has an overall safety profile comparable to INSURE or ETT administration for most procedural outcomes. In RCTs, the absolute risk of adverse events with LISA was generally low. Apnea and bradycardia occurred in fewer than 10% of infants, while cough/gagging, regurgitation, and multiple attempts occurred in approximately 10%–20% of cases. Hypoxemia was the most frequent event, affecting about one‐third of infants undergoing LISA. Non‐randomized studies showed similar patterns but higher absolute rates for some outcomes, particularly apnea and multiple attempts, suggesting variability in real‐world practice. When comparing LISA with INSURE or ETT, no statistically significant differences were observed for apnea, bradycardia (both author‐defined and heart rate < 100 bpm), hypoxemia, cough or gagging, or the need for multiple catheterization attempts. The only exception was surfactant regurgitation, which occurred more often with LISA.

The safety profile of LISA should be interpreted within in the context of neonatal practice. The absence of significant differences across most safety outcomes suggests that LISA does not introduce additional procedural risk compared to INSURE or ETT. The increased incidence of surfactant regurgitation with LISA is clinically relevant, as reflux may reduce the effective dose delivered and necessitate repeat administration. This phenomenon may reflect the mechanics of catheter placement and the gradual instillation technique inherent to LISA, which can predispose to retrograde flow if positioning is suboptimal. However, Kanmaz et al. (2013), indicates that reflux typically involves only a small fraction of the administered dose and does not compromise therapeutic efficacy, as oxygen requirements decreased even in infants with visible reflux [21]. In the arm‐level pooled analysis, hypoxemia emerged as the most frequent adverse event. These episodes are generally brief and reflect predictable physiological responses to airway instrumentation, including transient interruption of respiratory support and the inherent vulnerability of preterm infants to oxygen instability [55, 56]. Evidence indicates that such desaturation events are typically transient and can be managed effectively with minor adjustments in respiratory support [57]; however, their high prevalence highlight the need for vigilant monitoring throughout the procedure.

When comparing arm‐level incidence estimates across study designs, findings from non‐randomized were numerically higher for several outcomes, particularly apnea (16.0% vs. 2.7%) and the need for multiple attempts to place the catheter (24.5% vs. 17.4%). These differences likely reflect real‐world variability in technique, operator experience, and patient selection, which are less controlled than in randomized settings. Such factors can increase the likelihood of event detection and contribute to the moderate to high heterogeneity observed in several pooled estimates. Despite these differences, the overall patterns were aligned across study designs, supporting the conclusion that the relative safety profile of LISA remains stable across clinical contexts. Non‐randomized evidence provides important value for adverse event evaluation because it captures routine clinical practice and includes broader patient populations. This perspective helps quantify the true clinical burden of these events.

The subgroup analyses suggest that most safety outcomes remain stable across clinical scenarios, yet certain patient and procedural factors influence the likelihood of specific events. The increased regurgitation observed in infants at or below 34 weeks likely reflects procedural factors specific to thin‐catheter administration, including catheter position, catheter size and surfactant instillation technique [58, 59]. Catheter size may influence respiratory physiology during LISA, particularly in extremely preterm infants [60]. A similar pattern in studies without premedication or sedation indicates that infant movement and intact airway reflexes can contribute to catheter instability [58].

This study has several important strengths. It includes the most comprehensive synthesis to date of procedural safety outcomes associated with LISA and draws from both randomized and non‐randomized evidence. The use of pairwise comparisons and arm‐level pooled estimates allowed for complementary evaluation of relative and absolute risks. The inclusion of subgroup analyses also enabled exploration of clinically relevant modifiers, providing a nuanced understanding of how patient characteristics and procedural practices influence safety. Several limitations should also be considered. Safety outcomes were inconsistently defined across studies, and timing of event assessment varied, which contributed to moderate to high heterogeneity in pooled estimates. Because some studies assessed events during laryngoscopy, catheter or ETT placement, or surfactant administration, whereas others used less clearly defined immediate post‐procedure windows, event rates may have differed according to the procedural phase and duration of observation. This variability is particularly relevant for transient events such as apnea, bradycardia, and hypoxemia/desaturation. Some outcomes were reported in a limited number of studies, restricting precision and the ability to evaluate publication bias. Bradycardia outcomes were not uniformly defined across studies. Although we analyzed threshold‐based bradycardia, heart rate < 100 bpm, separately from author‐defined bradycardia, the author‐defined outcome reflects study‐specific criteria and should therefore be interpreted cautiously. Additional limitations include potential biases related to open‐label trial designs, where knowledge of treatment assignment may influence event reporting, and measurement bias due to imprecise or individualized interpretation of adverse events. Publication bias is also possible, as studies with positive results are more likely to be published. Furthermore, the reporting of safety events for INSURE or ETT was sparse, likely to reflect differential focus on the newer intervention, and may have introduced detection bias. Non‐randomized studies, while valuable for capturing real‐world practice, have inherent limitations such as residual confounding and variability in procedural protocols. Despite these limitations, the overall patterns were consistent across study designs, analytic approaches and subgroup analysis, supporting the reliability and the robustness of our findings.

5. Conclusion

This meta‐analysis shows that LISA has a safety profile comparable to INSURE and ETT, with regurgitation as the only event consistently more frequent with the LISA technique. Most other procedural events occur at similar and generally low to moderate rates across methods. Absolute incidences were higher in observational studies, reflecting differences in outcome reporting, patient selection, and procedural variability rather than intrinsic risk. These findings support the safe use of LISA in clinical practice and suggest that standardized procedures, training, and technique optimization may further improve safety in real‐world settings.

Author Contributions

Xuezheng Sun: conceptualization, methodology, writing – review and editing, supervision. Shaked Yarza: writing – original draft. Yuting Kuang: methodology, formal analysis, visualization, data curation, writing – review and editing. Shruti Sharma: validation, supervision, writing – review and editing. Jenny Uyei: data curation, software, formal analysis, writing – review and editing. Subhan Khalid: software, data curation, formal analysis, writing – review and editing. Daniel Fuentes: conceptualization, writing – review and editing.

Ethics Statement

This study is a systematic review and meta‐analysis of previously published data.

Conflicts of Interest

X.S., S.Y., and D.F. are employees of Client company, which funded this research. Y.K., S.S., J.U. and S.K. are employees of IQVIA Inc. which provides consultancy services to pharmaceutical and biotech companies.

Supporting information

Supporting File

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.