Long-term Non-Invasive Ventilation in Infants: A Systematic Review and Meta-Analysis

Background The use of long-term non-invasive ventilation (NIV) to treat sleep and breathing disorders in children has increased substantially in the last decade; however, less data exist about its use in infants. Given that infants have distinct sleep and breathing patterns when compared to older children, the outcomes of infants on long-term NIV may differ as well. The aim of this study is to systematically review the use and outcomes of long-term NIV in infants. Methods Ovid Medline, Ovid Embase, CINAHL (via EbscoHOST), PubMed, and Wiley Cochrane Library were systematically searched from January 1990 to July 2017. Studies on infants using long-term NIV outside of an acute care setting were included. Data were extracted on study design, population characteristics, and NIV outcomes. Results A total of 327 studies were full-text reviewed, with final inclusion of 60. Studies were distributed across airway (40%), neuromuscular (28%), central nervous system (10%), cardio-respiratory (2%), and multiple (20%) disease categories. Of the 18 airway studies reporting on NIV outcomes, 13 (72%) reported improvements in respiratory parameters. Of the 12 neuromuscular studies exclusively on spinal muscular atrophy type 1 (SMA1), six (50%) reported decreased hospitalizations and nine (75%) reported on mortality outcomes. Risk of bias was moderate to serious, and quality of the evidence was low to very low for all studies. Most studies had an observational design with no control group, limiting the potential for a meta-analysis. Conclusion The outcomes reported in studies differed by the disease category being studied. Studies on airway conditions showed improvements in respiratory parameters for infants using NIV. Studies on neuromuscular disorder, which were almost exclusively on SMA1, reported decreased hospitalizations and prolonged survival. Overall, it appears that NIV is an effective long-term therapy for infants. However, the high risk of bias and low quality of the available evidence limited strong conclusions.

iNtRODUctiON Rationale Long-term non-invasive ventilation (NIV), defined as respiratory support delivered through an interface outside the airway, has become the treatment of choice for a number of chronic conditions resulting in respiratory insufficiency or sleep and breathing disorders in infants and children (1)(2)(3). These conditions include airway disorders, neuromuscular disorders (NMDs), and disorders of the central nervous system (CNS) (3)(4)(5)(6). The shift toward NIV therapies may have been driven by improvements in NIV technology, a greater emphasis on home-based care, and a growing acceptance of NIV as a viable long-term respiratory support (1,6,7). With the increasing number of infants and children living at home using NIV, understanding the benefits and risks of NIV is becoming important not only for specialists involved in starting this therapy but also for pediatricians and primary care physicians providing care to these children within the community and policy makers responsible for decisions about provision of healthcare resources. While there is a considerable body of work describing the use of long-term NIV, including continuous positive airway pressure (CPAP) and bi-level positive airway pressure (BPAP), in a broad range of pediatric populations, less is known about its use in infants (8)(9)(10). Without sufficient data to suggest otherwise, similar NIV treatment approaches are likely followed in both infants and older children, despite key physiological differences in sleep and breathing patterns in infancy. Both sleep and breathing processes are immature at birth and continue to develop through infancy, resulting in change in sleep patterns and breathing control that continue through early life (11). Sleep occupies a greater proportion of time in infants compared to older children (12), which makes infants more vulnerable to respiratory disorders that disrupt sleep. Immaturity of central respiratory centers in infants contributes to increased respiratory events and a greater variability in oxygen saturation, both of which may be important for the normal development of respiratory control (11,13). Since sleep and breathing processes differ by age, especially in early life, the type of respiratory and sleep disorders treated with NIV, the response to NIV treatment, and the outcomes for NIV may also differ in infants as compared to older children.
Most data available on long-term NIV use in infants is limited to single-center observational studies with relatively small sample sizes (8). Aggregation of the available data for combined data analysis will improve our understanding of the risks and benefits of NIV therapy in the infant population.

Objective
The objective of this systematic review is to summarize the available evidence on the use of long-term NIV for infants and to estimate effect sizes for specific sub-populations and clinical outcomes compared to alternative respiratory care strategies.

Research Question
Does the use of NIV, compared to supportive care, or invasive ventilation, improve clinical outcomes for infants under the age of 2 years with chronic conditions resulting in respiratory insufficiency or sleep and breathing disorders? metHODS

Study Design
This review was conducted using systematic review methodology.

Participants
The inclusion criteria for this systematic review were as follows: (1) infants, defined by the Public Health Agency of Canada as ages 0-24 months inclusive (14); (2) NIV use, defined as breathing support delivered from outside the airway; and (3) long-term NIV use, defined as greater than three months outside of an acute care setting. For studies that examined a broader age range, the mean age of NIV initiation had to be less than 24 months in order to be included in this review, or data had to be presented separately for infants. We did not place any restrictions on study design or outcome eligibility.

Systematic Review Protocol
The protocol for this systematic review was developed according to the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines (15). The full protocol has been registered in the PROSPERO database for international prospective reviews (16).

Search Strategy
This systematic review is an extension of a prior scoping review on long-term NIV in children (8). The scoping review search strategy, using Medical Subject Headings (MeSH) and free-text terms for "child" and "non-invasive ventilation," was developed for MEDLINE (Ovid) and adapted for subsequent electronic databases with the full protocol published elsewhere (17) [see Table 1   . These data included author's name, year of publication, country of publication, study design, sample size, age of NIV initiation, NIV type, primary underlying disease conditions, comorbidities, and primary and secondary outcome measures. One reviewer (PKB) extracted the data, and 20% of data extraction was verified by a second reviewer (MCC).

Risk of Bias
The Cochrane Risk of Bias in Non-Randomized Studies of Interventions (ROBINS-I) tool (18)

Synthesis of Results
Studies were grouped by disease category (airway, NMD, CNS, cardio-respiratory or multiple disorders) after the data collection stage, to allow for adequate pathophysiological comparisons. Within each disease category, studies were grouped based on primary disease conditions. We included studies with infants who had multiple disease conditions under one disease heading if >75% of the infant cohort had the same disease condition; otherwise these studies were included in the multiple disorders category. Primary and secondary outcomes were established after data collection, during synthesis of the data, based on the most common and clinically relevant outcomes reported in studies with the same disease condition. Primary outcomes were as follows: (1) objective changes in respiratory parameters, (2) discontinuation of NIV, (3) hospitalizations, and (4) mortality. Secondary outcomes were as follows: (1) improvements in underlying disease conditions, (2) improvements in growth parameters, (3) NIV facilitation of extubation, (4) predictors of NIV requirement, (5) NIV success/failure, (6) adherence to respiratory support, and (7) mask complications. Studies were included in the synthesis if they reported on at least one primary or secondary outcome. Continuous data were presented as a weighted mean (standard deviation) or median (interquartile range) where appropriate. Results were grouped and reported based on the primary underlying disease category being studied. Primary outcomes were reported in both tabular and narrative format, while secondary outcomes were only reported narratively.

Pierre Robin Sequence
Seven studies (7/60, 12%) reported on infants with Pierre Robin sequence (PRS) using long-term NIV ( Table 2). Four studies (4/7, 57%) reported on primary or secondary outcomes and were synthesized for this review (31,(34)(35)(36). A cohort study reported normalization of polygraphy parameters and gas exchange post-NIV initiation ( Table 3) (31). A case series reported a decrease in respiratory rates, statistically significant improvements in respiratory effort, and normalization of respiratory gases after administration of NIV therapy in infants with PRS (36). Two studies on 16 infants with PRS reported discontinuation from NIV in 11 (69%) infants because of improvements in respiratory parameters (31,36). Two studies comparing infants on NIV and invasive mechanical ventilation showed that the length of hospitalization were shorter for infants on NIV than for those receiving invasive mechanical ventilation via a tracheostomy (

35).
No studies addressed survival outcomes in infants with PRS using long-term NIV. Adherence of infants to NIV was reported as excellent, showing more than 8 hours of NIV use per day in two studies (31,36), with only a 1-2 week period required to adjust to the mask ventilation (31,35). An additional cohort study demonstrated that infants with PRS using NIV were 10.43 times more likely to progress to a surgical airway compared to infants who required less advanced respiratory supports such as prone positioning and a nasopharyngeal airway (34).

Laryngo-Tracheomalacia
All four studies (4/60, 7%) on infants with laryngo-tracheomalacia (LTM) using long-term NIV reported on primary or secondary outcomes and were synthesized in the review ( Table 2) (38)(39)(40)(41). Three studies (3/4, 75%) reported on changes in respiratory parameters ( Table 3) (38,39,41). A case-control study of 10 infants with LTM showed improvements in respiratory frequency and respiratory effort in infants using CPAP or BPAP compared to spontaneous breathing (38). Normalization of arterial oxygen saturations after NIV use was seen in two studies (39,41). NIV discontinuation was reported in two studies, with a combined discontinuation due to improvement rate of 81% (13/16 infants) (39,40). No studies examined hospitalization or mortality outcomes. Improvement in chest wall deformity after NIV use in three patients and normalization of weight in four patients was reported in one case-control study (39). The same study • Changes in respiratory parameters • Changes in respiratory parameters • Number of subjects on NIV        also reported an average NIV use per day of 10.2 hours/day in seven infants (50).

Central Hypoventilation Syndrome
There were six studies (6/60, 10%) on NIV use for infants with central hypoventilation syndrome (CHS) that reported primary or secondary outcomes, and all six were summarized ( Table 2) (60)(61)(62)(63)(64)(65). The diagnosis of CHS was confirmed clinically in two studies (61,65), via PHOX2B gene mutation analysis in three studies (60,62,64), and unreported in one study (63). NIV was used in conjunction with negative extra-thoracic pressure ventilation (VNEP) therapy in two studies: in one study, it was used as the primary therapy (65) and, in the second study, CPAP was used to relieve upper airway obstruction not resolved with VNEP (61). Improvements in respiratory parameters were reported in two studies: one showed the normalization of the partial pressure of carbon dioxide and resolution of pulmonary hypertension following the use of NIV (65) and the other study showed improvements in hypoventilation for 50% (3/6) of infants ( Table 3) (61).
One study with six infants reported NIV discontinuation in two infants (33%) because of improvements in respiratory parameters; the remaining four infants were using NIV only during sleep (61). One cohort study reported mortality outcomes and a higher hospitalization time for infants using invasive mechanical ventilation compared to NIV ( Table 4) (60). Two studies showed parent-reported improvements in growth and development after NIV initiation using the results of a parent questionnaire (61,63). An additional two studies reported pressure-related effects of mask use, which were predominantly skin breakdown and mid-face hypoplasia (64,65). One cross-sectional study showed that it took less than a week for five of the six infants to adjust to NIV (61). A control before-after study of infants using two BPAP ventilators showed comparable sleep and respiratory parameters with both ventilators, with the exception of a greater decrease in the maximum transcutaneous carbon dioxide with the intelligent volume-assured pressured support compared to a traditional BPAP ventilator (62).

Synthesized Findings
After examining studies for all disease categories and respective outcomes, only three studies on infants with SMA1 reporting mortality outcomes were eligible for meta-analysis (44,46,51). The results of meta-analysis showed that there was a statistically significant decrease in the relative risk of mortality in the NIV group compared to the supportive care group (Figure 2).

Risk of Bias and Quality assessment of Outcomes
Risk of bias ranged from moderate to severe in all studies synthesized in this review ( Table 5). Study design was the main contributor to the low quality assessment of the studies. Almost all the included studies had an observational study design, which contributed to confounding bias in participant selection and selected reporting of results. Grading of the quality of the evidence for outcomes such as changes in respiratory parameters, discontinuation of NIV, hospitalizations, and mortality showed that the quality of evidence ranged from low to very low for all studies ( Table 6).

Summary of main Findings
To our knowledge, this is the first systematic review on the use of long-term NIV in infants. We identified studies on a diverse range of airway conditions in which NIV therapy improved the results of polysomnographic and respiratory parameters. With data available for NMD and CNS disorders limited to SMA1 and CHS, extrapolation of NIV benefits to other NMD and CNS disorders in infants is challenging. Not all outcomes were studied in all disease categories; length of hospitalization was the focus in studies of PRS, while hospitalizations and mortality were the focus in studies of SMA, and respiratory events and NIV discontinuation in the remaining groups. The overall quality of evidence to support appropriate conclusions was low to very low for all studies included in this review.
There is a diverse range of airway disorders that may benefit from NIV therapy. Previous studies have identified many conditions that can predispose infants to upper airway obstruction, including craniofacial disorders, laryngeal disorders, and nasal obstruction (76). Similarly, in this review, we identified NIV use in a wide variety of diseases associated with compromised airway function, the most common being OSA, PRS, ALTE, infants at risk for SIDS, and LTM. The improvement in respiratory parameters reported in infants with airway disorders reflects an overall benefit from NIV therapy. In addition, the underlying airway conditions have potential for improvement, as seen with the infants discontinuing due to underlying improvements, so there may be less risk with NIV compared to invasive mechanical ventilation. Extrapolating these results to conditions with a similar pathophysiology, but for which there is no evidence for NIV use in the literature, may be reasonable given the diversity of disorders represented in the available evidence.    The meta-analysis shows that the relative risk of mortality is significantly lower in infants using NIV compared to infants on supportive care. This decrease may be attributed to prolonged survival in infants using long-term NIV compared to supportive care.
By contrast, extrapolation of outcomes for long-term NIV use in NMD and CNS disorders may be more challenging. The data relevant to long-term NIV use for NMD and CNS disorders are almost exclusively from two conditions: SMA1 and CHS. SMA1 is a progressively deteriorating disorder that is usually fatal during infancy. This contrasts with other NMD disorders presenting in infancy, such as congenital myopathy and congenital muscular dystrophy, which may have a better prognosis or steadier course (7,58). The difference in prognoses of these conditions makes generalizing outcomes for NIV use in SMA1 to other NMD less appropriate. Similarly, CHS was the only CNS disorder for which data on long-term NIV use was available. NIV may be useful for other CNS disorders with accompanying respiratory compromise, such as congenital or acquired brain injury. Given the potentially unique physiology of CHS extrapolating the outcomes of NIV use for infants with CHS to other CNS conditions with different underlying respiratory pathophysiology may not be appropriate. Creation of national disease registries for infants and children using NIV will provide the opportunity to aggregate data on rare or minimally studied diseases and examine the use and outcomes of long-term NIV in these populations.
The outcomes that were reported in studies differed depending on the primary underlying disease category that was being examined. Studies of airway conditions predominantly reported on changes in respiratory parameters reported via polysomnography results and discontinuation of NIV. In addition, most studies reported short-term overnight polysomnography results; only one study had data on polysomnography results after long-term follow-up periods of NIV use in infants (29). Only one study on upper airway disorders reported on mortality outcomes (27) and none on hospitalization outcomes. Long-term outcomes, such as hospitalizations, intercurrent illness, growth and development, and quality of life warrant further study. Interestingly, studies on SMA1 predominantly reported on mortality and hospitalization outcomes, with only one study reporting on changes in respiratory parameters.
While the overall quality of the evidence available for the use of long-term NIV in infants is low to very low, there is a body of evidence that may help guide clinical practice. The reason for the low quality of the evidence included the study design and a high risk of bias due to the lack of blinding and randomization, and control for confounding variables. While these findings highlight the need for future studies of strong design and lower risk of bias, the available data still provide important information to inform treatment decisions for conditions where long-term NIV is being considered.

Limitations of the included Studies
We identified a number of research gaps present in the studies included within this review. There was only one study that compared the efficacy of CPAP and BPAP ventilation in a cohort of infants (38). Similarly, while some studies reported mask complications (9,21), only one compared the efficacy and practicality of different infant NIV masks (74). Only single studies were identified on the use of long-term NIV for infants with breath holding (42) and cardiac disease (66). Additionally, there were no studies on the clinical supports necessary for infants to be placed on NIV. It is important to know whether infants receive consultation and support from physicians, registered nurses, home care support, or a combination thereof, to determine whether a multidisciplinary NIV care plan is necessary for this population. The lack of comparison groups and/or homogeneity of outcomes reported precluded meta-analysis for most topics.
Additional issues relevant to long-term NIV use in infants that are not addressed in the current literature include: limitations in availability of masks and headgear; limitations in the availability of BPAP machines that are sufficiently sensitive to detect flow rates; the impact of NIV use on craniofacial growth and the impact of craniofacial growth on NIV use; comorbidities in infants using NIV; the impact of NIV on somatic growth and psychomotor development; and, most importantly, the impact of NIV use on quality of life for both infants and caregivers.

Limitations of the Review
Our review relied on the search methods and primary-level screening decisions of a scoping review on NIV in children with subsequent development of the research questions on NIV in infants. The methods to identify studies for the scoping review, however, were sufficiently inclusive to capture all relevant evidence on NIV in infants. We defined NIV for the scoping review on long-term NIV as breathing support outside the airway via an interface, consistent with the MeSH Long-term NIV in Infants Frontiers in Pediatrics | www.frontiersin.org February 2018 | Volume 6 | Article 13 terminology for NIV and, therefore, included CPAP as well as BPAP. Some investigators, however, do not consider CPAP as a mode of NIV because it requires spontaneous breathing from the patient (1,77). To address this concern, we reported the different ventilation types used by infants in the tables included in this review. Finally, we defined infants as ages 0-2 years based on the Public Health Agency of Canada definition (14).
Some investigators may not agree with this definition, as the Centre for Disease Control defines infants as less than one year of age (78). Regardless of the definition used, it is still unclear whether there are differences in the outcomes of pediatric NIV with respect to age. Future work should consider whether infants represent a distinct group within children using longterm NIV.

Long-term NIV in Infants
Frontiers in Pediatrics | www.frontiersin.org February 2018 | Volume 6 | Article 13 cONcLUSiON This systematic review examines the use and outcomes of long-term NIV in infants across a range of respiratory and sleep disorders. Improvements in respiratory parameters and discontinuation from NIV due to improvement in underlying conditions have been shown for a broad range of upper airway disorders, such as OSA, PRS, and LTM, in infants. Long-term NIV use in infants with SMA1 decreased hospitalizations and prolonged survival compared to infants on supportive care. Infants with CHS may also show improvements in respiratory parameters after using NIV and potentially avoid tracheostomy. NIV appears to be a feasible method of providing long-term respiratory support for infants with a wide range of underlying conditions; however, several methodological weaknesses limit any strong categorical conclusions. The findings of this systematic review are relevant to a broad range of stakeholders and can be used to help guide clinicians on the use of long-term NIV in infants.

aUtHOR cONtRiBUtiONS
PB conceptualized and designed the review, assessed articles for inclusion, extracted and analyzed data, interpreted the data, drafted the initial manuscript, and completed all subsequent revisions until submission. MC conceptualized and designed the review, assessed articles for inclusion, verified data extraction, and critically reviewed the manuscript. RF developed the search strategy, carried out the literature searches, and critically reviewed the manuscript. MA and BA assessed articles for inclusion, and critically reviewed the manuscript. AK provided guidance on study design and critically reviewed the manuscript. CF provided guidance on study design and review methodology and critically reviewed the manuscript. JM conceptualized and designed the review, assessed articles for inclusion, verified data extraction, interpreted the data, and critically reviewed the manuscript. All authors reviewed the manuscript and approved the final manuscript for submission.

acKNOwLeDgmeNtS
The authors thank Dr. Meghan Sebastianski (Knowledge Translation Platform, Alberta SPOR SUPPORT Unit, University of Alberta) for her expertise and guidance on systematic review methodology.

FUNDiNg
The authors have no financial relationships relevant to this article to disclose. PB received salary support through a Pediatric Non-Invasive Ventilation studentship from the Department of Pediatrics, University of Alberta. The project was supported by a mini-grant from the Respiratory Health Strategic Clinical Network of Alberta Health Services.