In total, 715 patients with a diagnosis of bronchiolitis were admitted to the seven PICUs during the six seasons considered. The seasonal admission numbers and percentages were relatively consistent across all seasons except for the 2020-2021 COVID-19 season, during which there was a 10-fold reduction (Figure 1, panel A). Excluding 13 patients from the 2020-2021 season, the remaining 702 patients were grouped according to the period of admission, with 451 patients admitted to PICU in the Pre-COVID period and 251 in the Post-COVID period. The temporal distribution of admissions is depicted in Figure 1 (Figure 1, panel B).
Epidemiology and severity at admission
In the Post-COVID period, children admitted to the PICU were generally older (73 days [IQR 40-178] vs 55 days [IQR 33-125], p=0.039) and presented with a significantly worse PIM-3 score at admission (median 0.52% [IQR 0.38-0.67] vs 0.46% [IQR 0.37-0.62], p = 0.046). They also had more comorbidities (21% vs 13%, p = 0.039): chronic respiratory conditions (4.4%), congenital cardiac disease (3.7%), neurologic (3.1%) and syndromic disorders (3.1%) were the most common underlying conditions in both periods (Table 1). RSV cases made up 82% of the sample and did not change over time. Bacterial infection and coinfection with two microorganisms occurred in 7% and 15% of the overall sample, respectively. The PICU LOS was not significantly different in the two periods considered (median 4.0 days [IQR 3.0-7.0] vs 5 days [IQR 3.0-8.0], p = 0.373). Sepsis occurred in 1% of cases and overall mortality was 0.3%.
Table 1 Characteristics of the sample by period.
Characteristic
|
N
|
Pre COVID,
N = 451
|
Post COVID,
N = 251
|
Overall,
N = 702
|
p-value
|
Large center
|
|
246 (55%)
|
127 (51%)
|
373 (53%)
|
0.300
|
Gender (M)
|
702
|
234 (52%)
|
147 (59%)
|
381 (54%)
|
0.144
|
Age (d), median(IQR)
|
702
|
55 (33, 125)
|
73 (40, 178)
|
62 (34, 142)
|
0.039
|
Ethnicity
|
702
|
|
|
|
0.236
|
Caucasian
|
|
343 (76%)
|
180 (72%)
|
523 (75%)
|
|
Arabic
|
|
55 (12%)
|
29 (12%)
|
84 (12%)
|
|
African
|
|
20 (4.4%)
|
17 (6.8%)
|
37 (5.3%)
|
|
Asiatic
|
|
22 (4.9%)
|
12 (4.8%)
|
34 (4.8%)
|
|
Hispanic
|
|
8 (1.8%)
|
6 (2.4%)
|
14 (2.0%)
|
|
Mixed ethnicity
|
|
3 (0.7%)
|
6 (2.4%)
|
9 (1.3%)
|
|
Afro-american
|
|
0 (0%)
|
1 (0.4%)
|
1 (0.1%)
|
|
Weight (kg), median(IQR)
|
700
|
4.60 (3.70, 6.30)
|
5.00 (4.00, 6.95)
|
4.80 (3.80, 6.50)
|
0.056
|
Comorbidities
|
702
|
59 (13%)
|
52 (21%)
|
111 (16%)
|
0.039
|
Ex-premature
Respiratory
Cardiac
Neurologic
Syndromic
|
702
702
702
702
702
|
61 (14%)
18 (4%)
17 (3.8%)
17 (3.8%)
13 (2.9%)
|
20 (8.0%)
13 (5.2%)
9 (3.6%)
5 (2.0%)
9 (3.6%)
|
81 (12%)
31 (4.4%)
26 (3.7%)
22 (3.1%)
22 (3.1%)
|
0.056
0.5
>0.9
0.2
0.6
|
PIM 3 score (%), median (IQR)
|
673
|
0.46 (0.37, 0.62)
|
0.52 (0.38, 0.67)
|
0.48 (0.37, 0.63)
|
0.046
|
FiO2 at admission, median(IQR)
|
394
|
0.35 (0.30, 0.40)
|
0.40 (0.30, 0.45)
|
0.40 (0.30, 0.40)
|
0.052
|
Lactates at admission (mmol/L) median(IQR)
|
297
|
1.40 (1.00, 2.12)
|
1.38 (0.90, 2.20)
|
1.40 (1.00, 2.20)
|
0.325
|
RSV cases
|
538
|
287 (84%)
|
155 (79%)
|
442 (82%)
|
0.465
|
Sepsis
|
538
|
1 (0.2%)
|
3 (1%)
|
4 (1%)
|
0.187
|
Bacterial infection
|
538
|
16 (5%)
|
21 (11%)
|
37 (7%)
|
0.800
|
Co-infection (2 microorganisms isolated)
|
538
|
33 (10%)
|
48 (25%)
|
81 (15%)
|
0.600
|
Abbreviations: PIM, Pediatric Index of Mortality; FiO2, Fraction of Inspired Oxygen; RSV, respiratory syncytial virus. Column N reports the number of records available for each variable
Respiratory support and adjunctive therapies
Data on respiratory support are reported in Table 2. NIV was used in 70% of the patients and the overall rate of IMV (alone or before and/or after NIV) was 17%.
The use of HFNC and the need for NIV and its duration did not show significant changes between the two periods. Nasal prongs (44%) and nasal masks (22%) were the two interfaces more commonly used. During the Pre-COVID period, nasal prongs were more frequently used (50% vs 33%, p < 0.001), whereas nasal masks were more commonly used during the Post-COVID period (34% vs 16%, p < 0.001).
NIV modes also changed with time, with PSV or BiPAP being used more frequently in the Post-COVID era (55% vs 26%, p < 0.001), while CPAP and A-PCV were mainly used in the Pre-COVID period (p <0.05 in both cases). The rate of NIV failure and complications remained consistently lower than 10% and 1%, respectively.
The rate of IMV use was comparable between the two periods (15% vs 17%) and the extubation failure was very low overall (5%). The seasonal comparison of respiratory support (Figure 2) confirmed the same trends, with substantial stability in the need for NIV and IMV over time.
The use of HFOV, NMB and surfactant in intubated patients did not change in the two periods considered, nor did the rate of chest tube placement, which was always very low (0.9%).
Table 2 Comparison between respiratory support and adjunctive therapies in pre-COVID and post-COVID groups
Characteristic
|
N
|
Pre COVID,
N = 451
|
Post COVID,
N = 251
|
Overall,
N = 702
|
p-value
|
HFNC
|
702
|
222 (49%)
|
134 (53%)
|
356 (51%)
|
0.360
|
NIV
|
702
|
320 (71%)
|
174 (69%)
|
494 (70%)
|
0.672
|
IMV
|
702
|
78 (17%)
|
38 (15%)
|
116 (17%)
|
0.626
|
NIV ventilation modes:
|
|
|
|
|
|
PSV/BiPAP
|
469
|
73 (26%)
|
101 (55%)
|
174 (37%)
|
<0.001
|
CPAP
|
469
|
126 (44%)
|
56 (31%)
|
182 (39%)
|
0.012
|
A-PCV
|
469
|
85 (30%)
|
26 (14%)
|
111 (24%)
|
<0.001
|
NIV interfaces:
|
|
|
|
|
|
Nasal mask
|
702
|
70 (16%)
|
86 (34%)
|
156 (22%)
|
<0.001
|
Facial mask
|
702
|
18 (4.0%)
|
3 (1.2%)
|
21 (3.0%)
|
0.086
|
Nasal prongs
|
702
|
224 (50%)
|
84 (33%)
|
308 (44%)
|
<0.001
|
Helmet
|
702
|
47 (10%)
|
33 (13%)
|
80 (11%)
|
0.360
|
Full face
|
702
|
33 (7.3%)
|
7 (2.8%)
|
40 (5.7%)
|
0.031
|
NIV complications
|
297
|
1 (0.9%)
|
1 (0.5%)
|
2 (0.7%)
|
0.999
|
NIV failure
|
534
|
31 (8.8%)
|
11 (6%)
|
42 (7.9%)
|
0.400
|
Chest tube positioning
|
702
|
4 (0.9%)
|
2 (0.8%)
|
6 (0.9%)
|
0.999
|
HFOV
|
116
|
8 (10%)
|
1 (2.6%)
|
9 (7.8%)
|
0.360
|
NMB
|
115
|
24 (31%)
|
13 (35%)
|
37 (32%)
|
0.672
|
Surfactant
|
117
|
10 (13%)
|
2 (5.1%)
|
12 (10%)
|
0.400
|
Extubation failure
|
116
|
4 (3%)
|
2 (2%)
|
6 (5%)
|
0.999
|
NIV length (d), median(IQR)
|
567
|
3.00 (2.00, 5.00)
|
3.00 (2.00, 4.00)
|
3.00 (2.00, 5.00)
|
0.360
|
IMV length (d), median(IQR)
|
115
|
11 (7, 14)
|
8 (5, 13)
|
10 (6, 14)
|
0.125
|
Antibiotic therapy
|
648
|
246 (56%)
|
138 (65%)
|
384 (59%)
|
0.086
|
PICU LOS (d), median(IQR)
|
702
|
5.0 (3.0, 8.0)
|
4.0 (3.0, 7.0)
|
4.0 (3.0, 7.0)
|
0.373
|
Mortality
|
702
|
0 (0%)
|
2 (0.8%)
|
2 (0.3%)
|
0.260
|
Abbreviations: HFNC, High Flow Nasal Cannula; NIV, Non-Invasive Ventilation; IMV, Invasive Mechanical Ventilation; PICU, Pediatric Intensive Care Unit; PSV, Pressure Support Ventilation; BiPAP, Bilevel Positive Airway Pressure; CPAP, Continuous Positive Airway Pressure; A-PCV, Assisted Pressure Controlled Ventilation; HFOV, High-Frequency Oscillatory Ventilation; NMB, Neuromuscular Blockade; LOS, Length of Stay
Comparison between Centers
The clinical characteristics of patients admitted for bronchiolitis according to the Center are reported in Table I-supplementary. There was huge heterogeneity across Centers in terms of both case mix and management practices. Some PICUs admitted predominantly younger children (median 50 days, IQR 34-104) compared to others (median 125 days, IQR 35-347). The PIM-3 ranged from 0.16% (IQR 0.12-0.51) to 0.53% (IQR 0.45-0.68). Using the average risk profile for intubation, the probability of a child with bronchiolitis being intubated varied from 4.7% to 50% across Units. Even NIV rates, modalities and interfaces used varied widely.
Risk factors for intubation
Univariable logistic regression models assessed the association between mechanical ventilation and patient characteristics (Table II - Supplementary). These results showed that patients’ age, ex-prematurity and the use of facial masks were associated with a lower risk of intubation outcome. Conversely, the use of nasal masks, nasal prongs and helmets were risk factors for a subsequent IMV.
Considering the hierarchical structure of the data, the best multilevel regression model for intubation outcome (Figure 3) included study period, patient’s age and severity (PIM-3), PSV/BiPAP/A-PCV ventilation modes and the size of PICU (large centers vs small ones). In this model, the Pre-COVID period represents a protective factor towards the risk of IMV (OR 0.38, 95% CI 0.16-0.89). Furthermore, RSV infection was associated with a higher risk of mechanical ventilation (OR 2.49, 95% CI 1.10-5.63). PIM3 score, age, the use of A-PCV/BiPAP/PSV modes and the size of the Center were not associated with the risk of intubation. For this model, the Intraclass Correlation Coefficient (ICC) of 0.22 indicates that approximately 22% of the total variance is attributable to differences between Centers and the variables entered into the model explained 65% of the variance of the outcome (Nagelkerke R2 = 0.645). Moreover, the model showed good accuracy, with an AUC of 0.76 (95% IC 0.70-0.81).