Acute respiratory failure (ARF) is a common cause for admission to the pediatric intensive care unit (PICU), and between 35% and 64% of these patients require invasive mechanical ventilation (IMV).1,2 Recent studies have shown that between 25% and 34% of the patients with ARF require prolonged invasive mechanical ventilation (pIMV).3–5 This percentage has increased over the past decade following technological advances used in ventilatory support and overall intensive care.4

However, currently, the definition of pIMV in children is unclear, and there is no consensus among experts. Former studies have used arbitrary cutoff timeframes from 96h to 21 days.6–13 This lack of uniformity complicates the comparison and generalization of results across different studies. Therefore, it is crucial to establish a clear and agreed-upon definition of pIMV in the pediatric population to facilitate research and clinical decision-making.10

In addition, prolonged ventilation has been associated with specific risk factors in the pediatric population like age <12 months, weight <10kg, and PRISM scores >20, among others (6–8). These factors can help identify patients at higher risk of developing pIMV and guide proper management strategies.

pIMV has also been associated with worse clinical outcomes like longer lengths of stay, and more morbidity, and healthcare costs.4,14–17 These complications highlight the importance of identifying and predicting the risk of pIMV in the pediatric population to implement preventive interventions and optimize clinical management.

In this context, the objective of this study is to describe a cohort of pediatric patients ventilated due to ARF from the Latin American Collaborative Network (LARed network) registry18 and, using this data, create a definition of pIMV based on statistical frequency (75th percentile). Afterwards, the goal is to determine whether this definition is associated with risk factors and subsequent events during the evolution and stay at the pediatric intensive care unit, as described in the scientific medical literature available. Additionally, a comparison will be drawn with cutoff timeframes described by other authors to assess the adequacy of our observation.

Retrospective analysis of a cohort with data collected prospectively from the ARF registry of the LARed Network (including nearly 40 hospitals in Latin America).18 All patients who received IMV during their hospital stay from April 2017 through January 2022 were studied.

A total of 8001 patients with ARF were included in the LARed database from April 2017 through January 2022, 1698 of whom were included for analysis (Fig. 1). The mean duration of mechanical ventilation in the entire cohort was 7 +/- 7 days, with a 75th percentile of 9 days. This classified 382 patients (22.5%) on pIMV. Both the distribution and overall survival curve are available in the supplementary data (Figs. 1S, 2S in Supplementary material).

Figure 1.

Diagram of patients on invasive mechanical ventilation admitted to the study.

(0.19MB).

The main outcome of the analysis in this contemporary multicenter cohort of patients on IMV was the identification of 9 days as a differential cutoff timeframe that contributes to the definition of prolonged invasive mechanical ventilation (pIMV) based on relevant outcomes in the routine clinical practice: a higher percentage of multiorgan dysfunction, presence of acute respiratory distress syndrome (ARDS), higher rate of healthcare-associated infections, and increased use of adjuvant therapies like neuromuscular blockade, antibiotics, prone position, and HFOV in those with on IMV for more than 9 days.

In pediatrics, there is no consensus on pIMV, and evidence for an appropriate definition is scarce. This knowledge gap leads to significant variability in the interpretation of results across different studies. The most commonly used definitions in the medical literature available of pIMV include the requirement of invasive mechanical ventilation for over 72h,19–21 7 days,8,22–26, 14 days,27–29 and up to 21 days.6,30,31 In this study of the LARed cohort, we used a different approach. pIMV was defined as a duration >9 days based on the 75th percentile of the dispersion statistic including 22.5% of the cases in this group.

The frequency of pIMV in the cohort studied is difficult to compare to other reports due to differences in definitions and patient heterogeneity.32 In our cohort, 73.6% of the patients were on IMV longer for over 72h, 36% for over 7 days, 8.5% for over 14 days, and 3.9% for over 21 days, similar data to the findings from other cohorts of patients with ARF.22–24,27–31 Similarly, we obtained a median course of ventilation of 5.23 days (2.92–8.62 days), similar to the group with mild ARDS from the PARDIE study with a median course of ventilation of 5.9 days (4.0−10.2 days).33

However, in this homogeneous cohort of patients with acute respiratory failure, we found that age <6 months and the presence of comorbidities, specifically bronchopulmonary dysplasia, were early risk factors of pIMV. Other studies in general ICU populations have also described that breastfed babies, premature infants or the presence of comorbidities are associated with pIMV (6–8). Interestingly enough, our result showed a negative correlation between asthma as the primary diagnosis leading to IMV and pIMV. The duration of ventilatory support was shorter in patients with asthma compared to those with bronchiolitis and pneumonia, which is similar to the findings reported in other cohort.34–37 This could be explained by the underlying pathophysiological mechanism, which often tends to be reversed within a short period of time in most patients with appropriate therap.37–39

It has been reported that bacterial coinfection in patients with respiratory syncytial virus infection leads to a longer course of mechanical ventilation.40 In our study, we did not find an association between bacterial or viral etiology of infection and pIMV. However, the decision to initiate antibiotic therapy was associated with pIMV in the univariate analysis, which could be considered as an indirect marker of sepsis or infection even in the absence of a positive culture. The diagnosis of fungal infection did show a strong association with pIMV.

The correlation between high scores in these severity scales as predictors of prolonged ventilation is controversial. Payen et al. found an association between the PRISM score and pIMV.7 In our study, we used the PIM3 scale and did not find any associations with pIMV, which is indicative that the most severely ill patient at the time of admission does not necessarily need to spend more time on IMV.

Respiratory system dysfunction occurred more frequently in pIMV with a higher number of cases of ARDS and greater need for ventilatory support. However, only a mean airway pressure >13cm H2O had a significant association in the multivariate analysis. Unlike other series, we found no association between pIMV and the degree of hypoxemia, but rather with the use of therapies like neuromuscular blockade and high-frequency ventilation. The pIMV group required more blood products, which is similar to what has been described in a general PICU cohort by Monteverde et al.6

In the multivariate analysis, we found an association between pIMV and healthcare-associated infections, particularly ventilator-associated pneumonia (VAP). This has also been reported in the general PICU population and in post-cardiac surgery patient.6 Other studies have demonstrated higher mortality rates in patients with prolonged mechanical ventilation beyond the usual average duration. Mortality rate associated with pIMV was 7%, which was not significantly higher compared to that reported in the cohort. However, pIMV was associated with longer lengths of stay and need for tracheostomy. Patients with pIMV had a lower Functional Status Scale (FSS) score when they were discharged from the PICU. Functional status scales have not been reported in previous pIMV cohorts. However, they are clinically important outcomes that assess not only acute-phase mortality but also the effect on the patients’ quality of life.

Our study has several limitations. Firstly, due to the lack of a consensus definition for prolonged mechanical ventilation in pediatrics, we chose to use a statistical definition based on the distribution of our population. While we acknowledge that this selection may be considered arbitrary, it allowed us to accurately characterize a significant subgroup of children with prolonged ventilation and obtain clinically relevant results. However, we should mention that the applicability of these findings to other series or cohorts can be limited, and further validation through replication of our findings in other cohorts is required. Secondly, a key limitation of the study is the lack of access to the temporary nature of events beyond the 9-day mark on invasive mechanical ventilation due to the way the database was built. As a result, we were unable to establish a causal order with the available data. Thirdly, our study is retrospective and based on the secondary analysis of a cohort of patients with ARF, which limits the analysis to the variables present in the database. In addition, the duration of mechanical ventilation and other outcomes may be influenced by the different medical practice of the different centers involved. To mitigate this, we used an adjusted mixed logistic regression model. A fourth significant limitation of our study is the lack of clinical information when patients were intubated. Including this data would have allowed a more precise study of the association between prolonged ventilation and other risk factors and clinical outcomes. However, due to the way the information was collected in the database, we only had access to data on the admission of patients to the pediatric intensive care unit. Finally, we should mention that the association of factors does not imply causality. Therefore, our results should be interpreted with caution and not used to draw definitive conclusions on the association between prolonged ventilation and admission risk factors or subsequent events in patients with ARF.

In our study, we saw that certain factors at admission, during disease progression, and during the stay at the PICU) are associated with prolonged mechanical ventilation, defined as a duration >75th percentile of the population (9 days). The results obtained in this study would allow us to intervene in modifiable risk factors and promote earlier weaning from mechanical ventilation to prevent complications and negative outcomes. Our data can be used to propose a definition of prolonged mechanical ventilation and serve as a metric of strategies to improve quality healthcare. Further research is needed to confirm these relations and, particularly, to establish whether there is an association between prolonged mechanical ventilation and the development of new morbidities.

Juan Sebastian Barajas-Romero, Pablo Vásquez-Hoyos, Sebastian Gonzalez-Dambrauskas, Franco Díaz, and Pietro Pietroboni participated in the study idea and design.

Pablo Vásquez-Hoyos, Rosalba Pardo, Juan Camilo Jaramillo-Bustamante, Regina Grigolli, Nicolas Monteverde-Fernández, Roberto Jabornisky, Pablo Cruces, Adriana Wegner, and Pietro Pietroboni participated in data curation.

Juan Sebastian Barajas-Romero, Pablo Vásquez-Hoyos, Sebastian Gonzalez-Dambrauskas, and Franco Díaz were responsible for the analysis and interpretation of data and drafting of the manuscript.

All authors contributed to the critical review of the intellectual content and final approval of the manuscript version to be published.