Use and outcomes of long-term noninvasive ventilation for infants

ABSTRACT

OBJECTIVE: The aim of this study is to compare the clinical characteristics, technology and outcomes between infants and older children using long-term NIV.

METHODS: In this 10-year retrospective review, 120 infants were matched to 240 older children in a 1:2 ratio based on sex and closest date of NIV initiation. Medical charts and sleep laboratory records were reviewed to extract demographic, NIV technology, polysomnography and clinic outcome data.

RESULTS: The results demonstrate a greater proportion of cardiorespiratory disease [16% vs. 6%, OR 3.04 (95% CI 1.47 to 6.31)] and a lower proportion of upper airway disorders [46% vs. 60%, OR 0.56 (95% CI 0.36 to 0.87)] in infants compared to older children. Infants had more comorbidities [4.0 (IQR 3.0) vs. 3.0 (IQR 2.0), p < 0.001] and used more additional technology [36% vs. 16%, OR 2.88 (1.73 to 4.78)] than older children. Improvements in respiratory parameters and NIV adherence were similar between groups. While NIV clinic discharge rates were similar, the reason for discharge differed with infants primarily ceasing NIV due to improvements in the underlying disease condition [42% vs. 30%, OR 2.04 (1.04 to 4.03)] or switching to invasive mechanical ventilation [10% vs. 1%, OR 11.43 (1.34 to 97.47)] while older children transferred to other services [9% vs. 35%, OR 0.17 (0.06 to 0.46)].

CONCLUSIONS: These results suggest infants are a distinct group with respect to NIV therapy and support the need for modifications in the approach to long-term NIV use in infants.

RÉSUMÉ

OBJECTIF: Le but de cette étude est de comparer les caractéristiques, la technologie et les résultats obtenus pour des nourrissons ayant recours à la ventilation non invasive (VNI) à long terme à ceux obtenus pour des enfants plus âgés.

MÉTHODES: Dans cet examen rétrospectif portant sur 10 ans, 120 nourrissons ont été jumelés à 240 enfants plus âgés selon un ratio de 1 : 2 en fonction de leur sexe et de la date la plus récente à laquelle la VNI a débuté. Les dossiers médicaux et les registres de laboratoires du sommeil ont été examinés pour en tirer des données démographiques, ainsi que des données relatives à la technologie de VIN, à la polysomnographie et aux résultats cliniques.

RéSULTATS: Les résultats démontrent une plus grande proportion de maladies cardiorespiratoires [16 % comparativement à 6 % RC 3,04 (IC 95 % 1,47 à 6,31)] et une plus faible proportion de maladies des voies aÉriennes supérieures [46 % comparativement à 60 %, RC 0,56 % (IC 95 % 0,36 à 0,87)] chez les nourrissons comparativement aux enfants plus âgés. Les nourrissons avaient davantage de comorbidités [4,0 (IIQ 3,0) comparativement à 3,0 (IIQ 2,0), p < 0,001] et faisaient davantage usage de technologie supplémentaire [36 % comparativement à 16 %, RC 2,88 (1,73 à 4,78)] que les enfants plus âgés. Les améliorations dans les paramètres respiratoires et l'adhésion à la VIN étaient similaires dans les deux groupes. Bien que les taux de sortie de la clinique étaient similaires, la raison pour la sortie différait : les nourrissons cessaient la VIN principalement en raison d'améliorations dans la maladie sous-jacente [42 % comparativement à 30 %, RC 2,04 (1,04 à 4,03)] ou du passage à la ventilation mécanique invasive [10 % comparativement à 1 %, RC 11,43 (1,34 à 97,47)], tandis que les enfants plus âgés étaient transférés dans d'autres services [9 % comparativement à 35 %, RC 0,17 (0,06 à 0.46)].

CONCLUSIONS: Ces résultats semblent indiquer que les nourrissons constituent un groupe distinct en ce qui a trait au traitement par VIN et viennent appuyer la nécessité d'apporter des modifications dans la façon d'aborder l'utilisation de la VIN à long terme chez les nourrissons.

KEYWORDS:

• Continuous positive airway pressure
• CPAP
• bilevel positive airway pressure
• BPAP
• sleep apnea syndromes
• respiratory insufficiency

Introduction

The number of children requiring long-term respiratory support has increased greatly over the last two decades,¹ at least in part, due to improvements in the survival of children with complex medical illness.^2,3 In addition, there has been a shift from using long-term invasive mechanical ventilation (IMV) to noninvasive ventilation (NIV) therapies, such as continuous positive airway pressure (CPAP) and bilevel positive airway pressure (BPAP).^4,5 The use of NIV, a method of delivering positive pressure breathing support via an interface outside the airway, has become the standard of care for providing long-term respiratory support for infants and older children with sleep and breathing disorders.^5,6 Reasons for the shift towards NIV therapies include improvements in ventilator technology,^4,6 avoidance of complications associated with IMV^7,8 and a growing recognition by clinicians and families of NIV as a viable alternative respiratory therapy.^4–6

Although the use of NIV for all children with sleep and breathing disorders has been increasing, there is less data on its use and outcomes in infants.^9–11 The available data suggests that the current treatment approach to NIV therapy is similar in infants and older children. However, there are key differences in the sleep and breathing physiology in infants that may impact their response to NIV.^12,13 For example, both total sleep duration and the proportion of total sleep time spent in rapid eye movement (REM) sleep is higher in early life and decreases across infancy and childhood.^12,13 Arousals are common after birth and decrease across the first year of life.^13,14 The central control of breathing is immature at birth and matures in the months following birth, resulting in an initial high variability in the respirator pattern in infants.^13,15 Respiratory events, including both central and obstructive events, commonly occur even in otherwise healthy infants and decrease with age.^16,17 These key differences in sleep and breathing in early life contribute to an increase in the risk for consequences of breathing disruption, and potential differences in response to respiratory support in infants compared to older children.

If the use and outcomes differ between infants and older children, a distinct approach to the use of long-term NIV for infants may be needed. Recognizing distinct benefits for long-term NIV in infants would help tailor its use as well as the support to families using this therapy. The aim of this study is to compare the clinical characteristics, technology and outcomes between infants and older children using long-term NIV. We hypothesize that the unique sleep and breathing physiology in infants will contribute to differences in the underlying disease conditions, higher levels of adherence and greater improvements in the underlying condition in infants using long-term NIV when compared to older children. Preliminary results were published in abstract form.¹⁸

Materials and methods

Study design and participants

This is a 10-year retrospective study comparing long-term NIV use for infants and older children. It is part of a larger study of all 622 children started on long-term NIV in the province of Alberta from January 2005 through December 2014.¹⁹ In the current study, all infants from the larger cohort were identified for inclusion. Infants were defined as age < 2 years and older children as age 2–18 years based on guidelines from the Public Health Agency of Canada.²⁰ Long-term NIV was defined as use for a minimum of three months outside of an acute care setting. Infants were matched to older children in a 1:2 ratio based on sex and the closest date of NIV initiation (within a 6 month period) to ensure contemporary comparisons. The study protocol was approved by Health Research Ethics Board for both participating institutions.

Data collection

Data were collected from the medical charts and sleep laboratory records of all children using long-term NIV at the two children's hospitals in the province of Alberta. These two institutions house the only publicly funded pediatric sleep laboratories in the province and, therefore, represent the majority, if not all, children using long-term NIV in the province. Data were collected at baseline and two follow-up times (6–12 months after initiation and the most recent clinic visit) from medical and sleep laboratory charts. Data were entered and stored in an electronic REDCap research database.²¹

Demographics collected from medical charts included age, sex, ethnicity, primary disease condition leading to the need for long-term NIV, comorbidities (defined as any chronic diagnosis aside from the primary underlying condition) and history of previous upper airway surgery. The primary disease conditions were grouped into five broad disease categories: upper airway, central nervous system, musculoskeletal and neuromuscular, cardio-respiratory (excludes upper airway) and unclassified conditions. Data collected on NIV technology included location of NIV start, NIV type, mask type, NIV pressures, adherence rates from NIV machine downloads (average hours of NIV used per night and % of days with NIV use for > 4 hours) and use of any additional technology. Additional outcomes included continuation of NIV, clinic discharge, reasons for clinic discharge (improvements in the underlying condition, patient/family decision to stop NIV, or transfer of services including transfer to adult services, another NIV provider or out-of-province care) and mortality.

Sleep studies included diagnostic polysomnography (PSG) studies, titration studies and split-night studies (including a diagnostic and titration component) and were performed using standard laboratory procedures. Data collection of PSGs was completed by an experienced technician following standard laboratory protocols. Studies collected before 2007 were sleep staged according to Anders et al²² (infants <6 months of age) or Rechtschaffen and Kales²³ (≥6 months of age) with scoring of respiratory variables based on the American Thoracic Society and previously published data in children.^24,25 The guidelines of the American Academy of Sleep Medicine²⁶ were applied to scoring starting in 2007. Apnea-hypopnea index (AHI) was calculated using the number of apneas and hypopneas that occurred during sleep divided by the total sleep time (TST). Obstructive AHI (OAHI) was defined as the sum of the obstructive apneas and obstructive hypopneas divided by the TST^26,27; OAHI >3 events/h in infants ≤6 months and >1 event/h >6 months of age was consider abnormal. Variables collected from sleep study charts included total sleep time, sleep efficiency (%), % time spent in REM sleep, arousal index, total AHI, OAHI, central AHI, % of total sleep time spent with oxygen saturations <90% and % of total sleep time spent with transcutaneous carbon dioxide levels >50 mmHg. To compare changes in sleep and respiratory parameters over time, parameters were compared between a diagnostic and titration PSG (full or split night studies) at baseline or between a baseline diagnostic and the first titration PSG after 3 months of NIV use. A minimum total sleep time of 120 minutes was required for both study types for inclusion in the analysis.

Analysis

Analyses were performed using IBM SPSS Statistics version 24.0 (SPSS, Inc., Chicago, IL). Summary statistics were reported as median (IQR) or frequencies (n, %) as appropriate. Chi-square test or Fisher's exact test, and Mann-Whitney U test were used to compare categorical and continuous variables between groups respectively. Wilcoxon Signed Ranks test was used to compare sleep and respiratory parameters between paired diagnostic and titration PSG studies. Odds ratios with 95% confidence interval (CI) or Mann-Whitney U test statistic were calculated to determine a measure of association for two dichotomous or continuous variables respectively.

Logistic regression was used for multivariable analysis of outcomes including continuation of ventilation, or clinic discharge due to improvement of the underlying condition, patient/family decision to stop NIV, transfer of services, switch to IMV, and mortality. Variables for inclusion were determined based on mechanistic plausibility in addition to demonstration of univariable association with the outcomes; these included demographics, sleep and respiratory parameters, NIV technology, and adherence variables. The 10-year study period was divided into three equal epochs to assess timing of NIV start in the regression model. Since our study is a comparison between infants and older children, age at NIV initiation was included in all models. A p-value of <0.05 was used to support statistically significant effects.

Results

The full data set included 622 children, of which 122 (20%) were infants; 120 infants were matched to 240 older children for the purpose of this study, with insufficient appropriate matches for 2 infants. The median age of NIV initiation in infants and older children was 9.0 (IQR 12.0 months) and 108.5 (IQR 91.8 months) respectively (Table 1). The primary disease category leading to the need for long-term NIV differed between groups, with older children having more upper airway conditions and infants having more cardiopulmonary disease. Infants also had a higher number of comorbidities compared to older children.

Table 1. Comparison of baseline clinical characteristics, including respiratory parameters, for infants and older children using long-term noninvasive ventilation.

Characteristics	Infants (n = 120)	Older children (n = 240)	Odds ratio (OR 95% CI) or Mann-Whitney U test statistic
Age (months), median (IQR)	9.0 (12.0)	108.5 (91.8)
Males (n, %)	68 (57)	136 (57)
Underlying disease category (n, %)
Upper Airway	55 (46)	144 (60)	0.56 (0.36 to 0.87)^*
Central Nervous System	26 (22)	38 (16)	1.47 (0.84 to 2.57)
Musculoskeletal/Neuromuscular	18 (15)	40 (17)	0.88 (0.48 to 1.62)
Cardiorespiratory	19 (16)	14 (6)	3.04 (1.47 to 6.31)^†
Other	2 (2)	2 (2)	1.00 (0.18 to 5.54)
Average number of comorbidities, median (IQR)	4.0 (3.0)	3.0 (2.0)	18749.50^‡
Comorbidity distribution (n, %)
0 co-morbidities	3 (3)	28 (12)	0.19 (0.06 to 0.65)^†
1–2 comorbidities	46 (38)	117 (49)	0.65 (0.42 to 1.02)
3–4 comorbidities	35 (29)	62 (26)	1.18 (0.73 to 1.93)
5+ comorbidities	36 (30)	33 (14)	2.69 (1.57 to 4.60)^‡
Prior A/T (n, %)	31 (27)	125 (54)	0.31 (1.19 to 0.51)^‡
Baseline diagnostic PSG data, median (IQR)
Total sleep time (TST; min)	234 (168)	308 (222)	7637.50^†
Sleep efficiency (%)	83 (14)	86 (14)	8668.00
REM sleep (%)	23 (17)	15 (10)	13272.50^‡
Arousal index (events/hr)	10 (10)	8 (10)	9258.00
Total AHI (events/hr)	23 (28)	10 (17)	13039.50^‡
Obstructive AHI (events/hr)	16 (27)	8 (14)	12540.50^‡
Central AHI (events/hr)	2 (6)	1 (3)	11305.00^*
% TST with SpO2 < 90%	4 (18)	0.3 (5)	13740.00^‡
% TST with TcCO2 > 50 mmHg	1 (36)	0.0 (18)	8543.50

Abbreviations: AHI, apnea-hypopnea index; Auto-PAP, automatic positive airway pressure; A/T, adenoidectomy and/or tonsillectomy; BPAP, bilevel positive airway pressure; CI, confidence interval; CPAP, continuous positive airway pressure; IQR, interquartile range; OR, odds ratio; PSG, polysomnography; SpO₂, arterial oxygen saturation; TcCO₂, transcutaneous carbon dioxide saturation_; TST, total sleep time.

* p < 0.05;

^† p < 0.01;

^‡ p < 0.001.

Baseline sleep and respiratory parameters also differed between infants and older children. Infants had a lower median total sleep time compared to older children. As expected, infants spent a higher proportion of total sleep time in REM sleep, and had a high total AHI, OAHI and central AHI compared to older children.¹³ Infants also spent a higher amount of TST with oxygen saturations < 90%.

There were differences between both groups with respect to NIV technology use. Infants were more likely to start NIV in the hospital (Table 2). Infants were more likely to use a nasal interface, BPAP ventilation and require more additional technology compared to older children. Recommended CPAP pressures were lower in infants, but BPAP pressures were comparable between both groups at the baseline visit. Infants received more home-care support compared to older children [67% vs. 26%, OR: 5.92 (3.46 to 10.14)].

Table 2. Comparison of technology use for infants and older children using long-term noninvasive ventilation.

NIV technology	Infants	Older children	Odds ratio (OR 95% CI) or Mann-Whitney U test statistic
Location of NIV start (n, %)
Home (pre-titration)	20 (17)	64 (27)	0.55 (0.32 to 0.97)^*
Home (post-titration)	45 (38)	137 (57)	0.45 (0.29 to 0.71)^†
Hospital ward	27 (23)	17 (7)	3.83 (1.99 to 7.37)^‡
NICU/PICU	27 (23)	21 (9)	3.05 (1.64 to 5.66)^‡
Interface type (n, %)
Nasal	97 (95)	132 (61)	12.64 (4.94 to 32.32)^†
Full face	5 (5)	86 (39)	0.08 (0.03 to 0.20)^†
NIV type (n, %)
CPAP or Auto-PAP	76 (65)	194 (81)	0.44 (0.27 to 0.72)^†
BPAP	41 (35)	46 (19)	2.28 (1.38 to 3.74)^†
NIV pressures (cmH2O), median (IQR)^a
CPAP or Auto-PAP	6.5 (3)	8 (4)	3143.50^†
BPAP (IPAP)	13 (3)	15 (7)	116.00
BPAP (EPAP)	6 (2)	6 (4)	104.00
Use of one or more additional technology (n, %)	43 (36)	38 (16)	2.88 (1.73 to 4.78)^†
Types of additional technology used (n, %)	33 (28)	32 (13)	2.47 (1.43 to 4.26)^†
Gastrostomy tube	7 (6)	4 (2)	3.66 (1.05 to 12.74)^*
Nasogastric tube	6 (5)	1 (0.4)	12.58 (1.50 to 105.72)^†
Supplemental oxygen	0 (0)	2 (1)	1.50 (1.40 to 1.62)
Oral appliance	0 (0)	1 (0.4)	1.50 (1.40 to 1.62)
Cough assist

Abbreviations: Auto-PAP, automatic positive airway pressure; BPAP, bilevel positive airway pressure; CPAP, continuous positive airway pressure; EPAP, expiratory positive airway pressure; IPAP, inspiratory positive airway pressure; NICU, neonatal intensive care unit; NIV, noninvasive ventilation; PICU, pediatric intensive care unit.

^a Post-titration recommended pressures at baseline visit.

* p < 0.05;

^† p < 0.01;

^‡ p < 0.001.

Treatment response and outcomes after starting NIV therapy showed both similarities and differences between groups. The average time between diagnostic and treatment PSGs was lower for infants compared to older children [5.0 (IQR 5.5 mo) vs 10.0 (IQR 13.5 mo), Mann-Whitney U statistic: 9366.50]. Data for sleep and respiratory parameters from a diagnostic to titration PSG showed similar improvements for infants and older children (Table 3). This included a significant decrease in the total and obstructive AHI from diagnostic to titration study in both groups (Figure 1). Infants had a greater decrease in the total sleep time with oxygen saturations < 90%. Infants were less likely to have their OAHI normalize on titration PSG compared to older children [9% vs. 23%, OR: 0.33 (0.13 to 0.89)]. The average hours of NIV used per night was similar between groups for the first follow-up visit, but differed in the second visit, with infants using NIV for more hours than older children (Table 3). The percentage of days with NIV use for >4 hours/24 hours was similar for infants and older children at both of the two follow-up visits. Post-hoc analysis with subjects split into three groups (<1 y, 1–2 y, >2 y of age) showed the average hours of NIV use per night, and the percentage of days with NIV use >4 hours, were similar between all groups at the first and second follow-up visits (data not shown).

Table 3. Comparison of treatment response and outcomes for infants and older children using long-term noninvasive ventilation.

Treatment response and outcomes	Infants	Older children	Odds ratio (OR 95% CI) or Mann-Whitney U test statistic
PSG indices^a, median of differences (IQR)	Change from Dx to Tx PSG	Change from Dx to Tx PSG
Total sleep time (TST) (min)^b	−18 (200)	−62 (242)^*
Sleep efficiency (%)	−1 (22)	−1.5 (21)
REM sleep (%)	4 (18)	5 (19)
Arousal index (events/hr)	−0.5 (6)	−2 (8)
Total AHI (events/hr)	−6 (21)	−3 (12)
Obstructive AHI (events/hr)	−5 (19)	−3 (11)
Central AHI (events/hr)	0.0 (4)	0.0 (2)
% TST with SpO2 < 90%	−2 (11)	−0.1 (3)^†
% TST with TcCO2 > 50 mmHg	0.0 (12)	0.0 (7)
Adherence (average hours used/night)
First follow-up visit, median (IQR)	6 (7)	6 (5)	2744.00
Second follow-up visit, median (IQR)	8 (8)	6 (5)	1351.00^*
Adherence (% days with NIV use > 4hrs)
First follow-up visit, median (IQR)	59 (70)	71 (65)	3411.00
Second follow-up visit, median (IQR)	83 (73)	67 (66)	2005.50
NIV outcomes^c (n, %)
Continuation of NIV	47 (44)	120 (52)	0.72 (0.46 to 1.15)
Clinic discharge	59 (56)	109 (47)	1.38 (0.87 to 2.20)
Reasons for clinic discharge (n, %)
Improvement in underlying condition	25 (42)	27 (27)	2.04 (1.04 to 4.03)^*
Patient/family decision to stop NIV	18 (31)	31 (30)	1.01 (0.50 to 2.02)
Transfer of services	5 (9)	36 (35)	0.17 (0.06 to 0.46)^‡
Switch to IMV	6 (10)	1 (1)	11.43 (1.34 to 97.47)^†
Mortality, (n, %)	14 (12)	11 (5)	2.750 (1.21 to 6.26)^*

Abbreviations: AHI, apnea-hypopnea index; Dx, diagnostic; NIV, noninvasive ventilation; PSG, polysomnography; REM, rapid eye movement; SpO₂, arterial oxygen saturation; TcCO₂, transcutaneous carbon dioxide saturation; TST, total sleep time; Tx, titration.

^a At the first follow-up visit,

^b Includes both full and split night studies, therefore, compares study sample time.

^c excludes mortality.

* p < 0.05;

^† p < 0.01;

^‡ p < 0.001.

Figure 1. Change from a diagnostic to titration polysomnography (PSG) in (a) the total apnea-hypopnea index (AHI) in infants; (b) the total AHI in older children; (c) the obstructive AHI in infants; (d) the obstructive AHI in older children. There was a similar decrease in total AHI (p = 0.23) and obstructive AHI (p = 0.36) from a diagnostic to titration PSG for both infants and older children.

While clinic discharge rates were similar between infants and older children, the reasons for discharge differed between groups (Table 3). Infants predominantly discontinued because of improvements in the underlying condition or switch to IMV. By contrast, older children were primarily discharged from clinic because of transfer to other services. Mortality was higher in infants than older children. For those who discontinued long-term NIV during the follow-up period, duration of NIV use was 14.0 (IQR 20) months in infants and 22.5 (IQR 29) months in older children (Mann Whitney-U Statistic 5672.00, p < 0.01).

Covariates for multivariable analysis of the reasons for NIV clinic discharge and mortality included age, underlying disease category, number of comorbidities, total AHI, OAHI, central AHI, % TST with SpO₂ <90%, % TST with TcCO₂ > 50 mmHg, location of NIV start, NIV type, additional technology use and % of days with NIV use >4 hours (first and second follow-up visit). Increasing age and BPAP were significant predictors of continued use of NIV (Table 4). Younger age and CPAP ventilation were predictors of clinic discharge due to improvement in underlying conditions. Lower adherence at the second follow-up visit was predictive of patients/family decision to stop NIV. Increasing age and NIV initiation in a hospital setting were predictors of clinic discharge due to transfer of services. BPAP and use of a greater number of additional technologies were predictors of switch from NIV to IMV. Age was not a significant predictor of mortality; instead a higher number of comorbidities and a greater use of additional technology were independent predictors of mortality.

Table 4. Results of binomial logistic regression with continuation of noninvasive ventilation, reasons for clinic discharge, and mortality as outcome variables and population characteristics and technology use as predictors.

Variables	Covariates	B ± SE	Exp(B)	95% C.I. for Exp(B)
Continuation of NIV	Age (months)	0.008 ± 0.002	1.008^‡	1.004 to 1.011
	NIV type	−0.802 ± 0.314	0.448^*	0.243 to 0.829
	Constant	0.077 ± 0.362	1.080
Improvement in underlying condition	Age (months)	−0.009 ± 0.003	0.991^†	0.985 to 0.996
	NIV type	1.084 ± 0.466	2.958^*	1.186 to 7.375
	Constant	−1.915 ± 0.444	0.147^‡
Patient/family decision to stop NIV	Age (months)	0.004 ± 0.004	1.004	0.995 to 1.013
	Adherence^a	−0.035 ± 0.010	0.966^†	0.947 to 0.985
	Constant	−0.932 ± 0.629	0.394
Transfer of services	Age (months)	0.023 ± 0.004	1.023^‡	1.015 to 1.031
	Location of NIV start	−1.625 ± 0.701	0.197^*	0.050 to 0.778
	Constant	−2.284 ± 0.533	0.102^‡
Switch to IMV	Age (months)	−0.013 ± 0.008	0.987	0.971 to 1.003
	NIV type	2.643 ± 0.973	14.053^†	2.086 to 94.673
	Additional technology	2.558 ± 0.971	12.907^†	1.924 to 86.594
	Constant	−4.475 ± 1.014	0.011^‡
Mortality	Age (months)	0.000 ± 0.004	1.000	0.993 to 1.007
	No. of comorbidities	0.222 ± 0.100	1.249^*	1.027 to 1.519
	Additional technology	1.748 ± 0.507	5.742^†	2.127 to 15.498
	Constant	−4.336 ± 0.657	0.013^‡

Abbreviations: IMV, invasive mechanical ventilation; NIV, noninvasive ventilation.

^a % of days with NIV use >4 hrs – second follow-up visit.

* p < 0.05;

^† p < 0.01;

^‡ p < 0.001.

Discussion

To our knowledge, this is the first study comparing the use and outcomes of long-term NIV between infants and older children. Infants had more comorbidities and required more additional technology than older children, potentially making them a more medically complex group. Changes in most of the sleep and respiratory parameters were similar between groups highlighting the efficacy and viability of NIV as a method of providing long-term breathing support across childhood. Despite similar incidence of central nervous system disease and musculoskeletal/neuromuscular disease, the use of BPAP was more common in the infant population. Reasons for clinic discharge differed, demonstrating that infants may have a wider spectrum of outcomes, including improvement or switch to IMV. Somewhat surprisingly, adherence was similar for both infants and older children.

Differences in underlying clinical characteristics and technology use between infants and older children support the idea that infants represent a more medically complex group that may require separate treatment approaches. Medically complex children have been defined as those having multisystem disease, functional impairment of daily living, complex medical regiments and/or technology dependence.^2,28 While our full cohort could be considered medically complex because of the need for long-term NIV, our results show that infants have a higher number of comorbidities and require more additional technology than older children. The impact of medical complexity on treatment strategies for children are not well understood, especially in infancy.^2,28,29 In our study, a higher number of comorbidities and greater use of additional technologies showed independent associations with mortality; this may simply be a marker of the underlying disease leading to the need for long-term NIV use. Further work is needed to determine appropriate treatment strategies to improve outcomes of long-term NIV use in infants which may include multidisciplinary care for high risk infants.

Improvements in sleep and respiratory parameters post-NIV initiation demonstrate that NIV is an effective therapy for improving sleep and breathing parameters in both infants and older children. Complications of disrupted sleep and breathing have been linked to abnormal growth, failure to thrive and cognitive impairment in infants;^30,31 and cardiovascular,^32,33 metabolic^34,35 and adverse neurobehavioral outcomes in older children.^36,37 Our analyses demonstrated improvements in sleep and respiratory parameters after starting NIV therapy, such as a higher percentage of REM sleep, decrease in the total and obstructive AHI and less total sleep time with oxygen saturations below 90%. These results are similar to those described in other infant studies reporting on improvement in PSG parameters with NIV for infants with obstructive sleep apnea and support benefit for NIV in other airway disorders.³⁸ With limited information on the impact of NIV on sleep and respiratory parameters outside upper airway disorders, additional work is needed to understand the relationship between improvements in sleep and respiratory parameters and subsequent health outcomes in other underlying disease categories for both infants and older children.

Although the most common type of NIV used in both groups was CPAP, more infants were using BPAP compared to older children. CPAP is commonly used in conditions where upper airway obstruction is present, while BPAP is beneficial for disorders resulting in abnormal central ventilatory drive or muscle weakness as it can help unload respiratory muscles and improve alveolar ventilation.^39,40 Since upper airway disorders occurred most frequently for both age groups in our population, the higher rates of CPAP use seems appropriate. With the incidence of CNS and NMD being similar for both infants and older children, it is interesting that the rates of BPAP use were significantly higher in infants. Lack of infant-specific guidelines around the use of long-term NIV may be a factor for clinicians deciding whether to initiate an infant on CPAP or BPAP. More infants start NIV in an acute care setting without a prior PSG and, therefore, may be initiated and subsequently continued on BPAP therapy after being transferred home. Additionally, BPAP, but not CPAP, is funded by the provincial health system in Alberta, suggesting funding could be a factor influencing treatment decisions. Infant-specific guidelines around the use of long-term NIV would provide a standard for decision making.

Contrary to our hypothesis, the adherence was similar between infants and older children, despite expected longer sleep time and, therefore, longer opportunity to use NIV in infants. Previous studies have established that adherence is a common problem with children using NIV therapy;^41–43 however, there are few studies focusing on challenges to NIV adherence in infants.⁴⁴ Removal of NIV masks during sleep has remained a problem in the older child and adults; however, this problem should be less in infants as many infants will lack the manual dexterity to remove the NIV mask. This should, theoretically, improve the total time infants are using NIV, and, therefore, adherence measurement. It is important moving forward to work with parents/caregivers and children to understand barriers to adherence to help improve NIV use and resulting outcomes in both the infant and older child population.

There were some limitations to our study that must be acknowledged. As this was a retrospective chart review, information was collected from the medical charts and, therefore, we were limited in the available outcome data. We chose to limit outcomes to those treatment responses as measured by polysomnography, adherence from NIV machine downloads and clinic discharge as objective outcomes. Infants and children who started on NIV near the end of our data collection period had a shorter length of follow-up so had less time to experience outcomes. Lastly, infants were defined as age 0–2 years based on a previously established definition.²⁰ This cut-off may not reflect an important physiological marker and, therefore, may not represent the ideal cut-off-point to define important changes for long-term NIV use.

Conclusions

Our results demonstrate that infants represent a distinct group within the overall pediatric NIV population. Infants differed from older children with respect to the underlying disease category necessitating NIV. Infants had more comorbidities, required more additional technology and used more BPAP than older children. While infants primarily discharged from NIV clinic because of improvements in underlying conditions, older children were mostly transferred to other services. Based on differences in underlying population characteristics, NIV technology and outcomes on NIV therapy, it is reasonable to conclude that infants represent a distinct group within the overall pediatric NIV cohort. These differences suggest a need for infant-specific treatment strategies and guidelines around the use of long-term NIV.

Declaration of interest

The authors have no potential conflicts of interest to declare.

Acknowledgments

The authors would like to thank Dr. Maryna Yaskina (Women and Children's Health Research Institute, University of Alberta, Edmonton, AB, Canada) for her guidance and expertise in statistical analysis. Preliminary results for this manuscript were presented at the American Thoracic Society Meeting 2017, Washington, DC, USA.

Additional information

Funding

P. K. Bedi received salary support through a Pediatric Noninvasive Ventilation studentship from the Department of Pediatrics, University of Alberta. The work was supported by a mini-grant from the Respiratory Health Strategic Clinical Network of Alberta Health Services.