Nasal Cannula with Long and Narrow Tubing for Non-Invasive Respiratory Support in Preterm Neonates: A Systematic Review and Meta-Analysis

Background: Cannulas with long and narrow tubing (CLNT) are increasingly being used as an interface for noninvasive respiratory support (NRS) in preterm neonates; however, their efficacy compared to commonly used nasal interfaces such as short binasal prongs (SBP) and nasal masks (NM) has not been widely studied. Material and Methods: Medline, Embase, CENTRAL, Health Technology Assessment Database, and Web of Science were searched for randomized clinical trials (RCTs) and observational studies investigating the efficacy of CLNT compared to SBP or NM in preterm neonates requiring NRS for primary respiratory and post-extubation support. A random-effects meta-analysis was used for data synthesis. Results: Three RCTs and three observational studies were included. Clinical benefit or harm could not be ruled out for the outcome of need for invasive mechanical ventilation (IMV) for CLNT versus SBP or NM [relative risk (RR) 1.37, 95% confidence interval (CI) 0.61–3.04, certainty of evidence (CoE) low]. The results were also inconclusive for the outcome of treatment failure [RR 1.20, 95% CI 0.48–3.01, CoE very low]. Oropharyngeal pressure transmission was possibly lower with CLNT compared to other interfaces [MD −1.84 cm H20, 95% CI −3.12 to −0.56, CoE very low]. Clinical benefit or harm could not be excluded with CLNT compared to SBP or NM for the outcomes of duration of IMV, nasal trauma, receipt of surfactant, air leak, and NRS duration. Conclusion: Very low to low CoE and statistically nonsignificant results for the clinical outcomes precluded us from making any reasonable conclusions; however, the use of CLNT as an NRS interface, compared to SBP or NM, possibly transmits lower oropharyngeal pressures. We suggest adequately powered multicentric RCTs to evaluate the efficacy of CLNT when compared to other interfaces.


Introduction
Respiratory distress is a common occurrence in the neonatal period which warrants intensive care admission, and an appropriate strategy for respiratory support is critical to improve neonatal outcomes. There is widespread use of noninvasive respiratory support (NRS) including high-flow nasal cannula (HFNC), nasal continuous positive airway pressure (nCPAP), biphasic CPAP (BiPAP), nasal intermittent positive pressure studies were also searched for possible inclusion. The literature search strategy for all the databases is provided in Supplement Table S1.

Inclusion Criteria
Randomized controlled trials (RCTs), quasi-RCTs, crossover trials, and observational studies were eligible for inclusion. Descriptive studies with no control arm, narrative reviews, case reports, case series, commentaries, and letters to editors were excluded.
Population: Preterm neonates (<37 weeks' gestation) requiring NRS modalities of NIPPV or NCPAP as primary or post-extubation support were included.
Intervention: CLNT Comparators: SBP or NM 2.3. Outcomes 2.3.1. Primary Outcome Treatment failure was defined as the need for escalation to a higher mode of NRS or the need for invasive mechanical ventilation (IMV) at any time within the first 7 days of initiating the support. The indication and criteria for escalation of respiratory support was as defined by the study authors. As treatment failure and the need for IMV are not mutually exclusive, we considered treatment failure and the need for IMV separately as primary outcomes.

Secondary Outcomes i.
Clinical outcomes: receipt of surfactant therapy, air leak (as defined by authors), nasal trauma occurring at any time until the discontinuation of respiratory support (any grade and severity, as defined by the authors), and duration of IMV and NRS (days). ii.
Surrogate outcomes: pressure transmission at the level of the pharynx or esophagus and the work of breathing (using an objective and validated scoring technique).

Data Extraction, Data Synthesis, and Quality of Evidence
Data extraction was performed by two authors independently using a prespecified structured proforma. R-software (version 3.6.2) (R Foundation for Statistical Computing, Vienna, Austria) was used for the meta-analysis [28]. The Mantel-Haenszel method and the inverse variance method for dichotomous outcomes and continuous outcomes were utilized, respectively. Between the studies, heterogeneity was evaluated using Cochran Q, I2, and τ2 values. The random-effects model was preferred over the fixed-effects model; this was owing to the fact that clinical heterogeneity was anticipated between the studies, as some of the studies had enrolled neonates in whom NRS was used as a primary modality, while others used a mixture of primary as well as post-extubation modality. Also, the type of respiratory support varied, with some utilizing CPAP and others NIPPV.
The overall effect estimate for each of the outcomes was expressed as risk ratios (RR) and risk difference (RD) for dichotomous outcomes and mean differences (MD) for continuous outcomes, with their 95% confidence interval (CI) depicted using forest plots. The within-group standard error of the mean (SEM) reported in a trial was converted to the corresponding standard deviation (SD) [29]. Publication bias was planned to be assessed by using funnel plots if we included 10 or more clinical trials in the meta-analysis.

Risk-of-Bias (RoB) Assessment
Cochrane RoB tool version 2.0 was utilized for RCTs [30] and Risk of Bias in Nonrandomized Studies of intervention (ROBINS-I) for non-RCTs [31]. Two authors (V.R. and A.R.) assessed the RoB independently, and conflicts were resolved by consensus and discussion.

Certainty of Evidence (CoE) Assessment
Grading of recommendations, assessment, development, and evaluations (GRADE) was used for assessing the CoE, which was rated as high, moderate, low, or very low [32][33][34]. The results of the meta-analysis were reported as per a modified GRADE working group recommendation [35] (Supplement Table S2).

Subgroup Analysis
The studies were pooled as subgroups depending on the type of NRS and the indication for NRS for investigating the heterogeneity: i. type of NRS: NCPAP vs. NIPPV ii. indication for NRS: primary, post-extubation, or both We intended to pool the results based on the gestational age of the neonates: <28 weeks and ≥28 weeks if two or more studies provided separate data for the relevant subgroup.

RoB of the Included Studies
Of the three RCTs included, one provoked concern due to issues in the domain randomization process (Gokce 2019), and the other two RCTs (Hochwald 2021, Maram 2019) had a low risk of overall bias [24][25][26]. For these RCTs, the randomization process was Web-based, deviations from intended interventions were minimal, the measured outcomes were objective, there was no missing data, and all the selected outcomes were reported.
Among the non-randomized studies, two (Sharma 2020 and Drescher 2018) had a serious risk of overall bias predominantly contributed by confounding, and one had moderate risk of bias in the domain of measurement of outcomes (Singh 2018) [18,23]. The risk of bias in the included studies is given in Supplement Table S4 (randomized trials) and  Supplement Table S5 (non-randomized trials).

Primary Outcomes Treatment Failure
Clinical benefit or harm could not be ruled out for the outcome of treatment failure between the two groups [RR 1.20, 95% CI 0.48-3.01; 3 studies, 521 participants]. The CoE was downgraded to very low certainty due to limitations in the study design, indirectness, inconsistency, and imprecision ( Figure 2, Table 2). The test for subgroup differences was not significant.

Need for IMV
The meta-analysis showed that clinical benefit or harm could not be ruled out with CLNT compared with NRS provided with SBP [RR 1.37, 95% CI 0.61-3.04; risk difference 23 more per 1000, 95% CI 20-125 more; 3 studies, 521 participants]. The CoE was rated down by two levels to low due to serious limitations of inconsistency and imprecision ( Figure 3, Table 2). The test for subgroup differences was not significant.

Need for IMV
The meta-analysis showed that clinical benefit or harm could not be ruled out with CLNT compared with NRS provided with SBP [RR 1.37, 95% CI 0.61-3.04; risk difference 23 more per 1000, 95% CI 20-125 more; 3 studies, 521 participants]. The CoE was rated down by two levels to low due to serious limitations of inconsistency and imprecision ( Figure 3, Table 2). The test for subgroup differences was not significant.

Nasal Trauma
Clinical benefit or harm could not be ruled out for the outcome of nasal trauma between the two groups [RR 0.49, 95% CI 0.21-1.11; 3 studies, 521 participants; CoE: low]. The test for subgroup differences for NRS modes, NCPAP and NIPPV, was significant. (Figure 3, Table 2).

Surfactant Treatment
Clinical benefit or harm could not be ruled out for the outcome of surfactant therapy between the two groups [RR 1.44, 95% CI 0.68-3.04; 2 studies, 292 participants; CoE: very low] (Figure 2, Table 2). The test for subgroup differences was significant.

Duration of IMV
The meta-analysis of 2 studies involving 292 participants could not provide any meaningful interpretation with the use of CLNT when compared to other interfaces [mean difference (MD) of 5.07 days and 95% CI of −1.04 to 11.19 days]. The CoE was rated down to very low certainty due to serious limitations in the study design, serious risk of bias, high heterogeneity, and imprecision ( Figure 3, Table 2). The test for subgroup differences was significantly different.

Duration of NRS
Clinical benefit or harm could not be ruled out for the outcome of duration of NRS between the two groups [MD 2.85 days, 95% CI −0.95 to 6.64; 3 studies, 521 participants], and CoE was very low (Figure 4, Table 2). ildren 2022, 9, x FOR PEER REVIEW 9 of 14 Figure 3. Forest plots depicting the effect estimates for the outcomes: air leak, nasal trauma, and invasive mechanical (IMV) duration.

Nasal Trauma
Clinical benefit or harm could not be ruled out for the outcome of nasal trauma between the two groups [RR 0.49, 95% CI 0.21-1.11; 3 studies, 521 participants; CoE: low]. The test for subgroup differences for NRS modes, NCPAP and NIPPV, was significant. (Figure 3, Table 2).

Duration of NRS
Clinical benefit or harm could not be ruled out for the outcome of duration of NRS between the two groups [MD 2.85 days, 95% CI −0.95 to 6.64; 3 studies, 521 participants], and CoE was very low (Figure 4, Table 2).

Air Leak
Clinical benefit or harm could not be ruled out for the outcome of air leak between the two groups [RR 1.20, 95% CI 0.36-4.00; 3 studies, 521 participants], and the CoE was low ( Figure 3, Table 2).

Primary Outcomes Primary Outcomes
No meta-analysis was performed, as only one study reported treatment failure between the groups. The treatment failure reported by Drescher et al. was not statistically different between the two groups [RR 2.81, 95% CI 0.84 to 9.36]
No meta-analysis was performed for other secondary outcomes, as only a single observational study reported on some of these outcomes. This study (Drescher 2018) provided data for nasal trauma, IMV duration, and NRS duration. The IMV duration was significantly higher with CLNT with a mean difference of 7.58 days and 95% CI of 0.32 to 14.84. The same study reported nasal trauma as one of the secondary outcomes, which was significantly lower with CLNT [RR of 0.16, 95%CI 0.05 to 0.49, CoE: very low]. Duration of NRS was significantly lower with CLNT with a mean difference of 8.70 days and 95% CI of −15.88 to −1.52.

Discussion
In this systematic review and meta-analysis, we evaluated the efficacy of CLNT in comparison to other routinely utilized nasal interfaces for NRS, such as SBP and NM. To the best of our knowledge, this is the only systematic review and meta-analysis highlighting the evidence from six available clinical studies, three randomized (n = 521) and three non-randomized (n = 138), evaluating the efficacy of CLNT as an interface for NRS in preterm neonates for primary respiratory support or for post-extubation. Though CLNT possibly results in lower oropharyngeal pressure transmission when compared to SBP/NM (meta-analysis of non-RCTs), clinical benefit or harm could not be ruled out for most of the outcomes evaluated, highlighting the need for future trials on this question. Our finding of lower oropharyngeal pressure transmission with CLNT is consistent with the earlier in vivo and bench studies on CLNT showing lower pressure delivery with CLNT [4,[15][16][17][19][20][21]36]. The RAM cannula is quite similar to the traditional nasal prongs used to provide supplemental oxygen. Its long and narrow tubing is supposed to be more effective than traditional cannula in transmitting pressures to the nasal end, which in turn is relatively soft compared to SBPs [24][25][26]. Due to the very low CoE, the findings of our meta-analysis need to be validated by future studies.
Of the primary outcomes evaluated, treatment failure and need for IMV were assessed separately, as these outcomes are not mutually exclusive and neonates who meet the criteria for treatment failure may be supported with a higher mode of NRS rather than being initiated on IMV. We found no significant differences for both outcomes; however, the confidence limits included a significant benefit and harm, thereby obviating the ability to draw any meaningful conclusions. The softer nasal cannula which comes in contact with the nares is perceived to be associated with lesser nasal trauma compared to SBPs or NM [24][25][26]. We evaluated nasal trauma as one of the secondary outcomes, and it is usually one of the determining factors for the preferential choice of a specific NRS interface in most of the neonatal units. Although our meta-analysis indicated no difference in nasal injury with the use of CLNT cannula, clinical benefit or harm could not be ruled out due to an imprecise effect estimate and low certainty of the evidence. The betweenstudies heterogeneity in this meta-analysis for the outcomes could possibly be explained by the differences in the patient population, co-interventions, baseline event rates, and the variable assessment methods and criteria used to evaluate the outcomes [24][25][26]. Other factors to be considered include the nurse-to-patient ratio, the skill and training of the healthcare personnel, and intensity of monitoring, all of which are known to influence the important outcomes.
Our study has several strengths. To date, to the best of our knowledge, this is the only systematic review and meta-analysis summarizing the current evidence on a pertinent question of clinical importance related to the use of an increasingly adopted NRS interface, CLNT, in preterm neonates for primary and post-extubation support. The study followed the standard methodology per the PRISMA framework, the Cochrane group guidance for systematic reviews including a comprehensive search strategy, prospective protocol registration in PROSPERO, and explicitly defined the clinical question and synthesized the evidence using appropriate methods. Lastly, our review also examines the CoE using the GRADE working group guidelines, an essential aspect to appraise the quality and strength of evidence and thereby making an informed clinical decision.
However, we acknowledge some of the limitations of this meta-analysis. Firstly, the number of studies contributing to the systematic review and meta-analysis is limited. Secondly, despite synthesizing the evidence in a meta-analysis, the confidence in the evidence for many outcomes studied is limited due to very low to low evidence certainty. We identified a few ongoing trials (CTRI NCT03121781, NCT0216825, and CTRI/2020/03/024097) that may likely address these limitations [40][41][42]. We also recognize that there was a lack of homogeneity among the studies about the type of NRS used as well as its indication of use (primary support versus post-extubation), comparator interfaces, varying definitions, and assessment of primary and secondary outcomes, as evident by statistical heterogeneity for many outcomes. We explored the heterogeneity by pooling the studies as subgroups based on the type and indication of NRS and found the test for subgroup differences was statistically significant for some outcomes suggesting one of possible reasons for the observed heterogeneity. We also attempted to study the subgroup data based on the gestational age of the included neonates: (<28 weeks versus >28 weeks). However, limited reporting in the included RCTs or observational studies precluded such analyses.

Conclusions
The results of this meta-analysis indicate that compared to SBP or NM, the use of CLNT as an interface for NRS either as a primary support modality or for post-extubation in preterm neonates possibly results in reduced oropharyngeal pressure transmission; however, the meta-analysis could not assess with certainty the effect on important clinical outcomes. Henceforth, we suggest this as priority research and until more high-quality evidence is available, clinicians should consider proven nasal interfaces, such as NM or SBP, that are likely to provide effective pressure transmission, adequate ventilation, and oxygenation, thereby possibly avoiding IMV in preterm neonates.