Introduction: A detailed understanding of respiratory support patterns in preterm infants is lacking. The aim was to explore and visualize this practice in Sweden. Methods: Preterm infants with gestational ages of 22–31 weeks, admitted to neonatal units reporting daily to the Swedish Neonatal Quality Register and discharged alive in November 2015–April 2022, were included in this descriptive cohort study. Proportions receiving mechanical ventilation, noninvasive support, or supplemental oxygen were calculated and graphically displayed for each gestational week and postnatal day (range 0–97) up to hospital discharge or 36 weeks of postmenstrual age. Results: Respiratory support in 148,515 days of care (3,368 infants; 54% males; median [interquartile range] birthweight = 1,215 [900–1,525] g) was evaluated. Trajectories showed distinct nonlinear patterns for each category of respiratory support, but differences in respiratory support over the gestational age range were linear: the proportion of infants on mechanical ventilation decreased by −11.7 to −7.3% (variability in estimates related to the postnatal day chosen for regression analysis) for each week higher gestational age (r = −0.99 to −0.87, p ≤ 0.001). The corresponding proportions of infants with supplemental oxygen decreased by −12.4% to −4.5% for each week higher gestational age (r = −0.98 to −0.94, p < 0.001). At 36 weeks of postmenstrual age, dependencies on mechanical ventilation, noninvasive support, and supplemental oxygen varied from 3%, 84%, and 94% at 22 weeks to 0%, 3%, and 5% at 31 weeks of gestational age, respectively. Conclusions: Respiratory support patterns in very preterm infants follow nonlinear, gestational age-specific postnatal trajectories in a dose-response-related fashion.

© 2022 The Author(s). Published by S. Karger AG, Basel

Introduction

Respiratory support is vital in the management of lung immaturity and respiratory failure affecting preterm infants. Adequate respiratory support promotes onset of breathing and transition at birth and mitigates respiratory distress syndrome [1-3]. The progress, type, and duration of lung disease, as well as the type of respiratory support used for treatment, aid decision-making about other interventions such as centralization of care [4], surfactant replacement therapy [5, 6], and use of postnatal corticosteroids [6]. In preterm infants <28 weeks of gestational age (GA), advanced respiratory support forms the backbone of infant survival [7, 8]. Prolonged respiratory support is a hallmark of bronchopulmonary dysplasia (BPD) [9, 10].

Although an important everyday practice, the evidence base is limited in the most immature infants [7] and detailed knowledge on respiratory support in preterm infants is lacking. Most previous reports have aggregated data on use of respiratory support at different postnatal ages (typically at 28 days of postnatal age or 36 weeks of postmenstrual age [PMA]) to predict survival or post-discharge respiratory morbidity, or to define BPD in preterm infants [9-17]. Smaller and older single-center studies have described the timeline for postnatal oxygen dependency and mechanical ventilation in some more detail [18, 19], but there is no information on how different modes of respiratory support are used on a daily basis in large and contemporary cohorts of preterm infants.

A careful mapping of respiratory support – from birth to discharge – would be a powerful way to fill this gap in knowledge. The aim of this exploratory study, without a predefined hypothesis to test, was to create a detailed gestational age-specific graphical presentation of current practice of respiratory support in very preterm infants. For this purpose, we mapped the use of mechanical ventilation, noninvasive respiratory support (NRS), and supplemental oxygen in almost 150,000 days of neonatal care in a nationwide cohort.

Methods

The Swedish Ethical Review Authority approved this study on December 22, 2021 (reference number 2021-05855-01). This authority also waived patient informed consent. All caregivers had been informed during their stay in the neonatal units that perinatal data were registered, with a possibility to opt out at any time.

All data were de-identified and originated from the Swedish Neonatal Quality Register (SNQ). The SNQ collected data on procedures and outcomes for all infants admitted for neonatal care. All 37 neonatal units in Sweden have reported to SNQ since 2012, and the completeness and validity of the register have been found to be excellent [20].

All liveborn infants ≥220days weeks of GA were registered in SNQ. In 2017–2021, there were 1,761 liveborn infants, and 1,730 neonatal unit admissions were registered at 22–27 weeks GA (31/1,761 = 1.8% delivery room deaths). Moreover, and according to open-access statistics from SNQ [21], the mortality rates for admitted infants born in 2017–2021 were 55% at 22 weeks’ GA, 34% at 23 weeks, 27% at 24 weeks, 16% at 25 weeks, 8% at 26 weeks, 7% at 27 weeks, and <5% at 28–31 weeks’ GA.

Until 2015, data were summarized at discharge and then reported to SNQ. In 2015, a web-based service for neonatal units to report data daily was introduced. In 2017, all regional level III-IV centers were connected, and in 2021, 34/37 neonatal units had embraced daily reporting which occurred in 15,264/15,993 (95.4%) admissions for neonatal care in Sweden, online supplement Figure S1 (see www.karger.com/doi/10.1159/000527641 for all online suppl. material).

For this study, all infants delivered at 22–31 week gestation, reported daily to SNQ, and discharged to home between November 2015 and April 2022 were included (N = 3,368). Based on previous work demonstrating a very skewed distribution of mortality across the GA range [22], and that half of those who died after births at 24–28 weeks GA did so in the first postnatal week [23] despite active management [7], we decided to exclude infants who died during neonatal care (n = 364). Infants with more than minor malformations as defined by European network of population-based registries for the epidemiological surveillance of congenital anomalies (EUROCAT) [24] were also excluded (n = 339). To estimate potential bias introduced by these exclusions, sensitivity analyses including infants who died and infants with malformations were performed.

Risk factors or exposures were GA at birth in completed weeks and postnatal age in days, from birth (day 0) to hospital discharge to home or discharge from hospital-based homecare programs, or after the infant had reached 36 weeks’ PMA, whichever came first. Outcomes were categorized as (a) any supplemental oxygen (but information on FiO2 was not collected on a daily basis in SNQ) via mask, hood, or low-flow nasal cannula; (b) NRS defined as use of nasal continuous positive airway pressure, high-flow nasal cannula, noninvasive positive pressure ventilation, or noninvasive neurally adjusted ventilatory assist; or (c) mechanical ventilation defined as any form of respiratory support delivered through a tracheal tube (but ventilator settings were not collected on a daily basis in SNQ). Proportions of infants without any respiratory support were also calculated.

Statistical Analyses

Each infant received a random 6-digit number in SNQ. This unique number allowed for de-identified analysis of distinct numbers of infants (denominator) and type of respiratory support (numerator) per postnatal day. Each category of respiratory support reported on the same day for the same infant was assigned to this day. For infants transferred to another hospital with the last day of the first admission overlapping with the first day of the second admission, only 1 day of admission was generated.

Numbers (proportions), mean (standard deviation), and median (interquartile range [IQR]) values were calculated. χ2 test, 95% confidence intervals (95% CIs), Mann-Whitney U and Kruskal-Wallis tests were used to test for group differences. Linear regression and Pearson’s correlations were used to assess relations up to 34 days of postnatal age and at 36 weeks’ PMA. After 34 days of postnatal age, infants born at the highest GA had reached the endpoint (36 weeks of PMA) of data collection. Imputation of missing data was not performed because the study was exploratory and descriptive (nonanalytical) by design. For the same reason, no sample size or power calculations were performed. A p value <0.05 or 95% CIs without any overlap were considered as indicators of a statistically significant difference. Calculations and graphics were performed using Microsoft Excel 365, SPSS version 26, and SAS version 9.4.

Results

In the study period, 5,183 infants <32 weeks of GA without malformations were admitted for neonatal care and discharged alive in Sweden, online supplement Figure S2. Included infants (with at least one admission reported daily to SNQ; N = 3,368) had significantly lower GA (median 29.3 [IQR 27.0–30.7] versus 29.9 [28.0–31.0] weeks, p < 0.001) and lower birth weight (1,215 [900–1,525] versus 1,301 [1,000–1,601] g, p < 0.001) than infants with aggregated data registered in SNQ after hospital discharge (N = 1,815). The sex distribution did not differ between the two groups (54.5 vs. 55.3% boys, p = 0.79). The included infants generated 148,515 days of care that could be assessed for respiratory support.

Patterns of Mechanical Ventilation

Mechanical ventilation was linearly and inversely associated with GA in a dose-response-related pattern over the full range of postnatal days. On postnatal day 0, the proportion mechanically ventilated varied from 5.8% at 31 weeks GA to 100% among survivors born at 22 weeks. On postnatal day 27, the corresponding proportions were 0.2% and 88%. The largest proportion of mechanically ventilated infants was seen on day 1 (940/3,209 = 29%), decreasing to 462/3,117 (15%) on day 6 and to 262/2,728 (9.6%) mechanically ventilated infants on day 27.

In infants ≥25 weeks’ GA, the median duration of mechanical ventilation was ≤2 days. In infants born at 24 weeks’ GA, the median (IQR) duration of mechanical ventilation was 22 (2–38) days, at 23 weeks it was 32 (20–43) days, and at 22 weeks’ GA the duration of mechanical ventilation was 54 (34–62) days. At 36 weeks PMA, ventilator dependency varied from 0.2% at 31 to 3.0% at 22 weeks’ GA (Fig. 1, 2; Table 1).

Table 1.

Proportions (%) of very preterm survivors with or without respiratory support by GA and postnatal age.

Fig. 1.

Proportions (%) of very preterm survivors with mechanical ventilation per postnatal day and stratified by gestational age in weeks (22–31) up to a postmenstrual age of 36 weeks (N = 3,368).

Fig. 2.

Forrest plot of linear regression β-coefficient for associations between gestational age and mechanical ventilation, noninvasive respiratory support*, oxygen treatment, or no respiratory support, respectively, by postnatal age in days. The β-coefficients indicate the change in proportion (%) of infants treated with mechanical ventilation, noninvasive respiratory support, supplemental oxygen, or no respiratory support by 1 week increase in gestational age. r = Pearson correlation coefficient. The associations between GA and noninvasive respiratory support were nonlinear (parabolic) before 36 weeks of postmenstrual age (see Fig. 3). Therefore, linear regression coefficients were not displayed for those associations.

Patterns of NRS

NRS was initially associated with GA in a nonlinear, parabolic pattern. In the neonatal period, the largest proportions of infants noninvasively supported were found at 26–27 weeks’ GA with decreasing rates at both lower (22–25-week infants mechanically ventilated) and higher GAs (28–31-week infants; without any mechanical respiratory support).

At 36 weeks PMA, NRS had become linearly and inversely associated with GA. The median (IQR) duration with NRS ranged from 2 (0–5) days at 31 weeks to >75 days at 25 weeks GA, and at 36 weeks PMA, 2 of 3 infants of 22–24 weeks GA were still dependent on NRS. There was no steady state in proportions treated with NRS at 36 weeks’ PMA (Fig. 2, 3; Table 1).

Fig. 3.

Proportions (%) of very preterm survivors with noninvasive respiratory support per postnatal day and stratified by gestational age in weeks (22–31) up to a postmenstrual age of 36 weeks (N = 3,368). Labels at the end of each curve denote proportions (%) treated at the study endpoint.

Patterns of Supplemental Oxygen Therapy

Use of supplemental oxygen was inversely associated with GA in a dose-response-related pattern over the full range of postnatal days. The regression coefficients at different postnatal days varied between −4.5% and −12.4% for each week higher GA (with r = −0.94 to −0.98 and p < 0.001 for all models).

In the first postnatal week, a decline in use of supplemental oxygen was seen with a GA-specific nadir. The decline was less pronounced (mean [95% CI]: −9.5 [0.9 to −19.9] versus −41 [−37 to −45]%) and lasted shorter (2.2 [1.0–3.4] versus 8.2 [6.7–9.7]days) in infants born at 22–27 than in infants born at 28–31 week’ GA.

After the initial decline, proportions treated with supplemental oxygen increased to a maximum on postnatal day 20 (1,381/2,854 = 48% treated, ranging from 9.0% at 31 weeks to 100% at 22 weeks’ GA). Supplemental oxygen at 36 weeks’ PMA varied from 5.8% at 31 weeks to 94% at 22 weeks’ GA. At 36 weeks’ PMA, use of supplemental oxygen had not reached steady state – a significantly larger proportion was treated with oxygen at 35 (721/2,488 = 29%) than at 36 weeks’ PMA (533/2,243 = 24%, p < 0.005). The median (IQR) postnatal days with supplemental oxygen varied from 1 (0–2) days at 31 weeks to >100 days at 22 weeks’ GA (Fig. 2, 4; Table 1).

Fig. 4.

Proportions (%) of very preterm survivors treated with extra oxygen per postnatal day and stratified by gestational age in weeks (22–31) up to a postmenstrual age of 36 weeks (N = 3,368). Labels at the end of each curve denote proportions (%) treated at the study endpoint.

Patterns of No Respiratory Support

Only 141/3,199 (4.4%) infants were managed without any respiratory support on postnatal day 0. At day 10, 90% at 31 weeks GA, 76% at 30 weeks, 49% at 29 weeks, and 20% at 28 weeks GA did no longer receive respiratory support whereas in infants <28 weeks’ GA, respiratory support was still almost universal. The median (IQR) postnatal age when the infants breathed room air without respiratory support varied from 3 (1–6) days at 31 weeks to >100 days at 22 weeks’ GA (Fig. 2, 5; Table 1).

Fig. 5.

Proportions (%) of very preterm survivors breathing air without any form of respiratory support per postnatal day and stratified by gestational age in weeks (22–31) up to postmenstrual age of 36 weeks (N = 3,368). Labels at the end of each curve denote proportions (%) treated at the study endpoint.

Sensitivity Analyses

Including infants with malformations and infants that died before discharge did not alter the trajectories of supplemental oxygen treatment by GA, online supplement Figures S3, S4.

Discussion

This study demonstrated distinct gestational and postnatal age-specific trajectories for respiratory support in very preterm infants. The findings have implications for the understanding of how lung disease after preterm birth evolves, for considerations of treatment recommendations based on use of respiratory support, and for diagnostic criteria in neonatal care. They also reflect what capacity, competences, and skills are needed at different GA and postnatal age. The graphs displaying postnatal trajectories of respiratory support at different GAs may be helpful for epidemiological, educational, and counselling purposes. The long and varying periods of mechanical ventilation in infants born <26 weeks’ GA stand out as particularly concerning for quality, safety, and comfort in neonatal care. Future research should explore this variability and aim at finding ways to reduce the exposure to mechanical ventilation in these patients.

Expected GA at birth is used as an indicator of the need for mechanical ventilation after birth, and an anticipated ventilator need is used to justify antenatal transfer of pregnant women at risk to a hospital with neonatal intensive care facilities. In Sweden, a national recommendation on antenatal transfer for extremely preterm births <28 weeks of GA was endorsed during the study period. Outside the 8 centers with level III-IV neonatal care, 22/29 (76%) hospitals with level II neonatal units accepted preterm deliveries at ≥28 weeks GA. This contrasts to other countries in which centralization of preterm deliveries has been recommended at <32 weeks’ GA [25]. Varying recommendations reflect differences in organizational conditions, such as capacity among level II units to provide advanced NRS at birth and short-term (≤2 days) mechanical ventilation [4]. However, different recommendations may also reflect that current needs of respiratory support at different GAs have previously been insufficiently clarified. In infants born at 28–31 weeks’ GA, 3 out of 4 did not need any mechanical ventilation and the median duration among those who did was short (≤2 days).

Ongoing mechanical ventilation after 1–2 weeks has been suggested as an indication for postnatal corticosteroids [6, 26]. The rates presented herein then suggest that 4 of 5 infants born at 22–24 weeks and 1 of 3 infants at 25 weeks GA would be eligible for postnatal corticosteroids. However, such practice raises concerns about adverse neurological outcomes and must be weighed against the potential benefits [27]. In European infants of 22–29 weeks’ GA, large regional variations in use of postnatal corticosteroids have been reported [7, 28] without a clear relation to neonatal characteristics, survival, or BPD [7, 28, 29].

Early respiratory support patterns have been identified as antecedents of BPD [30, 31]. A recent US study reported birth weight, GA, FiO2, and respiratory support on day 1 after birth as significant risk factors for death or BPD severity [17]. Using a regression model, which in addition included infant sex and exposure to antenatal corticosteroids, the authors of the US study developed a web-based “BPD Outcome Estimator” that provides estimates for the probabilities of death or BPD by severity grade not only on day 1, but also at postnatal days 3, 7, 14, and 28 [17] SNQ data have recently been used to identify risk factors for mortality in very preterm infants, and for prediction of death within the first hours after birth [32, 33]. Adding early respiratory support patterns on day 1 might refine these tools.

Oxygen dependency at 36 weeks PMA has been questioned as a definition of BPD because of inaccuracy related to varying indications (saturation targets) for oxygen therapy and poor predictive ability of post-neonatal lung function [9, 10, 34]. Our finding that neither oxygen dependency nor noninvasive support had reached a steady state at 36 weeks PMA most likely adds to the limited association between BPD and adverse lung function in childhood [10, 35]. Respiratory support at later time points (beyond 36 weeks’ PMA) has been suggested to improve prediction of post-prematurity respiratory disease [10, 36]. However, given that information on respiratory support after 36 weeks’ PMA may not be readily available, GA may be used as a predictor of long-term lung function and respiratory disease risk in people born very preterm [35, 37, 38].

The respiratory support provided to preterm infants is largely determined by unit practices and guidelines [7]. The SNQ provides annual reports on variation between regions/units and practice changes over time. In the latest report [21], duration of mechanical ventilation in 2019–2021 varied significantly by greater region: between 20 and 38 days in infants <25 weeks’ GA, between 9 and 13 days at 25–27 weeks, and between 3.5 and 6.7 days in infants with 28–31 week’ GA. Changes over time were also seen but were less pronounced, except for significantly decreasing duration of mechanical ventilation in infants of 28–31 week’ GA.

A limitation of this study was that not all hospitals in Sweden embraced daily reporting at the start of the study period. Second, the slightly lower GA and birth weight in hospitals reporting daily (included) than in those who did not (noneligible) may reflect some selection bias. The most likely explanation is that all level III neonatal units had changed to daily reporting to SNQ before all level I-II units were connected, favoring inclusion of extremely preterm infants <28 weeks’ GA. Third, the study was right-censored and the trajectories of respiratory support after 36 weeks’ PMA – which have been discussed in relation to risk assessment of long-term outcome [36, 39] – were undisclosed. Fourth, the study did not include information on delivery room practices, fraction of inspired oxygen, oxygen saturation targets, type of mechanical ventilator or ventilator settings, permissive hypercapnia, caffeine, or weaning strategies. Fifth, other risk factors than gestational and postnatal ages were not evaluated at this stage. Sixth, the validity of the findings outside Sweden remains to be established.

The strengths and novelties of this study include exploration a unique set of data, allowing for detailed visualizations of respiratory supports used in very preterm infants. This is also the first time such data-driven information has been gathered from a whole country and the data were reported with a minimum of delay. The study was not left-censored on GA and included also infants born at 22–23 weeks of GA. Sensitivity analyses demonstrated robustness of the data and excluded potential bias introduced by excluding infants with malformations and infants who died.

Conclusion

In conclusion, respiratory support in very preterm infants follows nonlinear, GA-specific postnatal trajectories in dose-response-related patterns. Disclosing these patterns adds to a more detailed knowledge on how lung disease after preterm birth evolves, stimulates thoughts on future research, as well as considerations on treatment recommendations and diagnostic criteria. The graphs presented herein may also be helpful for educational and counselling purposes.

Acknowledgments

We thank all neonatal departments in Sweden for contributing and sharing data to our Neonatal Quality Register (SNQ).

Statement of Ethics

The Swedish Ethical Review Authority (Swedish Ethical Review Board) approved this study on December 22, 2021 (reference number 2021-05855-01). The authority mentioned above waived patient informed consent. All caregivers had been informed during their stay in the neonatal units that perinatal data were registered in the Swedish Neonatal Quality Register, with a possibility to opt out at any time.

Conflict of Interest Statement

The authors have no conflicts of interest to disclose.

Funding Sources

Mikael Norman was supported by a grant from a regional agreement on clinical research (ALF) between Region Stockholm and Karolinska Institutet (2020-0443) and by the Childhood Foundation of the Swedish Order of Freemasons, and Baldvin Jonsson by grants from a regional agreement on clinical research (ALF) between Region Stockholm and Karolinska Institutet (2020-0302), the Swedish Heart Lung Foundation (2016-027), and Region Stockholm clinical research appointment (DNR RS 2019-1140). The Swedish Neonatal Quality Register was funded by the Swedish Government (Ministry of Health and Social Affairs) and the body of Regional Health Care Providers. Role of the funders/sponsors: the funding bodies played no role in (1) study design; (2) the collection, analysis, and interpretation of data; (3) the writing of the report; and (4) the decision to submit the paper for publication.

Author Contributions

Stellan Håkansson and Mikael Norman had full access to the data and take responsibility for the integrity of the data. Concept and design; interpretation; and critical revision of the manuscript for important intellectual content: Mikael Norman, Baldvin Jonsson, Jonas Söderling, Lars J Björklund, and Stellan Håkansson. Acquisition and analysis of data: Stellan Håkansson, Jonas Söderling, and Mikael Norman. Drafting of the manuscript: Mikael Norman. Obtained funding: Mikael Norman and Baldvin Jonsson. All authors have approved the final version of the manuscript and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Data Availability Statement

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.

This article is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC). Usage and distribution for commercial purposes requires written permission. Drug Dosage: The authors and the publisher have exerted every effort to ensure that drug selection and dosage set forth in this text are in accord with current recommendations and practice at the time of publication. However, in view of ongoing research, changes in government regulations, and the constant flow of information relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for any changes in indications and dosage and for added warnings and precautions. This is particularly important when the recommended agent is a new and/or infrequently employed drug. Disclaimer: The statements, opinions and data contained in this publication are solely those of the individual authors and contributors and not of the publishers and the editor(s). The appearance of advertisements or/and product references in the publication is not a warranty, endorsement, or approval of the products or services advertised or of their effectiveness, quality or safety. The publisher and the editor(s) disclaim responsibility for any injury to persons or property resulting from any ideas, methods, instructions or products referred to in the content or advertisements.