(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Association between nirsevimab coverage and pediatric emergency department visits for bronchiolitis in 12 European countries: An interrupted time-series analysis [1]

['Léa Lenglart', 'Pediatric Emergency Department', 'Robert Debré University Hospital', 'Assistance Publique Hôpitaux De Paris', 'Paris Cité University', 'Paris', 'Université Paris Cité', 'Université Sorbonne Paris Nord', 'Inserm', 'Iame']

Date: 2026-09 

We used routinely collected data from 27 pediatric emergency departments across 12 European countries (01/2018–03/2024). All bronchiolitis cases aged <12 months were included. Regions were classified as intervention or control based on nirsevimab implementation. Coverage was obtained from official regional reports. We estimated changes in monthly case using controlled interrupted time-series models accounting for seasonality and temporal trends, by age-groups (0–3 and 3–12 months) and regions during the intervention period (10/2023–03/2024) and examined their correlation to coverage using Pearson’s test. Urinary tract infections served as a control outcome.

Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: Dr LL declares consulting activities for Biomerieux, Pr NO declares receiving travel grants from MSD, Pfizer, Sanofi, and GSK. None of these entities has a financial stake in the results of the present study. DB have been a co-investigator in trials addressing the role of long active monoclonals in RSV prevention. No other authors have conflicts of interest to disclose.

Data Availability: The minimal dataset supporting the conclusions of this article is publicly available on the following repository: https://doi.org/10.5281/zenodo.21625436 . The dataset includes aggregated data used for the analyses and figures. Nirsevimab coverage estimates were obtained from publicly available national or regional reports, which are cited in the manuscript.

However, despite very similar immunization guidelines, the recent literature reports substantial heterogeneity in the number of bronchiolitis cases averted across regions implementing nirsevimab [ 15 , 16 ]. This variability raises concerns about the overall public health impact of current strategies and, more importantly, about the factors driving these differences. Since RSV testing is not systematically performed in routine PED care across Europe, overall bronchiolitis cases represent a complete and pragmatic measure of the real-world population-level impact of RSV immunization. To better inform future policy decisions, addressing this question requires a multinational framework that enables the comparison of the diverse approaches.

The recent introduction of RSV immunization [ 4 , 5 ] is a revolution in the field of RSV prevention in infants. Indeed, in September 2023, nirsevimab was first implemented in 3 European countries: France, Spain and Luxembourg, in the prevention of RSV-bronchiolitis in infants. Since its first implementation, many studies demonstrated its real-world effectiveness in primary care visits [ 6 ], PEDs visits [ 7 – 9 ], inpatient admissions [ 10 – 12 ] and intensive care hospitalizations [ 13 , 14 ].

This retrospective observational study involving human participants followed the ethical standards of the Declaration of Helsinki. As this study used routinely collected, anonymized data and involved no direct interaction with participants, informed consent was not required and was therefore neither written nor oral. Following initial approval by the UK Health Research Authority (ID: 284008), all participating sites obtained ethical approval and data sharing agreements from their national/local institutional review boards for aggregate data.

Secondary analyses were performed between the overall coverage in infants aged under one year and the change in the number of hospitalized bronchiolitis, bronchiolitis discharged home from PEDs and in RSV positivity rates. All secondary outcomes were analyzed using the previously described methodology, however, for to assess the change in RSV positivity rates, we used quasi-binomial regression models with a logit link in the interrupted time series analysis to account for overdispersion in proportions and to constrain predicted values between 0 and 100%.

The following sensitivity analyses were performed: (i) to account for the heterogenicity in the number of patients included by each center, we performed a Pearson correlation analysis weighted on the number of cases included; (ii) to assess the influence of a potential non-independence between estimates, we conducted a sensitivity analysis excluding the control-country estimates from the Pearson correlation; (iii) in addition, we performed a meta-regression to account for between-center variability; (iv) to relax the assumption of linearity required by Pearson’s correlation, we performed a Spearman’s rank correlation; (v) to account for our number of correlation estimates being small, we performed a Kendall correlation test, which is more appropriate for small samples and non-parametric data; (vi) to assess whether the observed trends were dependent on the choice of the controls, we re-estimated the estimates using non-controlled interrupted time-series models and (vii) to account for overdispersion with a different variance structure than the quasi-Poisson model, we re-estimated the estimates using a negative binomial regression [ 27 ]. Two sensitivity analyses were added and performed during the reviewing process: (viii) to account for clustering of observations within control centers, we performed a mixed-effects negative binomial model with center-specific random intercepts; and (ix) heterogeneity in coding practices of bronchiolitis were accounted for by re-assessing the Pearson correlation using lower respiratory tract infections as the outcome.

To achieve this, we first assessed the evolution of monthly bronchiolitis cases for each of the 4 regions of the nirsevimab group using interrupted time series analysis models controlled on the evolution of bronchiolitis cases in the control group. Quasi-Poisson regression was used to account for overdispersion in the count data [ 25 – 27 ]. Seasonality was modeled with Fourier terms (sine and cosine functions with 12- and 6-month periods) [ 25 – 27 ]. The secular trend was modeled using a continuous time variable, and the number of bronchiolitis cases observed in the control group was included in the model as an external control [ 28 ]. One dummy variable (step-change) accounted for the intervention period (from October 2023 to March 2024) [ 25 , 26 ] and one continuous variable accounted for the time since intervention (slope-change). Moreover, another dummy variable accounted for the period of implementation of non-pharmaceutical interventions against SARS-CoV-2 (from April 2020 to March 2021). Model validity was assessed through visual inspection of residuals, autocorrelation functions, and Q-Q plots [ 27 ] (see Appendix E in S1 Text ). For each region of the nirsevimab group, and for the control group, we then calculated the overall change during the intervention period by estimating the fitted value of the number of observed cases compared to the expected value of the counterfactual scenario—if no intervention had occurred—based on the model parameters for each time point. The percentage of reductions and their 95% confidence intervals were obtained through parametric simulation (10,000 random draws from the normal distribution of model-fitted values for the counterfactual scenario).

The monthly number of urinary tract infection cases presenting to the PEDs and identified using ICD-10 codes (see Appendix B in S1 Text ) was used as a control outcome as it is not expected to be influenced by the implementation of nirsevimab or other seasonal drivers as bronchiolitis.

The main outcome of the study was the monthly number of bronchiolitis cases presenting to the PEDs over the study period in the 0–3 months and 3–12 months age-groups, assessed by controlled interrupted time-series analysis in the 4 regions from the nirsevimab group and the control group. These two age-groups were defined based on the hypothesis that nirsevimab implementation for infants born “in season” should drive 0–3 months bronchiolitis epidemiology, while nirsevimab implementation for infants born “out of season” should drive the 3–12 months bronchiolitis epidemiology (see Appendix D in S1 Text ). To note, two centers from the control group (POR002 and POR004) were excluded from the main analysis because of the unavailability of age categories data.

The guidelines regarding nirsevimab implementation in the Paris Region, Basque Country, Madrid Region and Catalonia are provided in the Table 1 . Based on those guidelines, two target populations were identified: infants born “in season”, meaning during the RSV season, for which the immunization was recommended at birth and infants born “out of season”, for which various catch-up immunization programs were implemented. We defined coverage as the number of infants receiving nirsevimab divided by the number of eligible infants. The coverage in each target population (born in-season or out-of-season) across regions was obtained from official reports by health agencies [ 20 , 21 ] or from published data [ 22 – 24 ].

We retrospectively collected monthly aggregated and anonymized data for each participating center through a clinical report form including: age category (0–3 months old = until 12-weeks old; and 3- to 12-months old); diagnosis (bronchiolitis or urinary tract infections; which served as a control outcome); and clinical management (discharged home or admission). As in previous works from this research group [ 17 , 18 ], the identification of patients was based on the ICD-10 coding system (see Appendix B in S1 Text ) or on the diagnosis provided by the PED physician in the health record of the case. Consistency in coding practices across and within study sites was monitored at the beginning of the study and overtime. The number of tests performed and RSV positivity rates were obtained through the laboratory databases of a subgroup of centers. The availability of data for each center is provided in Appendix C in S1 Text . All data were stored into secure and validated REDCap electronic data capture tools [ 19 ].

The results of the secondary analysis are presented in Table 3 and Appendix H to K in S1 Text . As observed for the primary outcome, the estimates of the reduction of the secondary outcomes varied across region of implementation. As an example, we observed a reduction in the RSV positivity rates during the intervention period, ranging from −35.2% (95% CI [−50.0, +37.5]) in the Paris region to −55.4% (95% CI [−156.7, +88.4]) in the Basque country, while it remained stable during this period in the control countries, +1% (95% CI [−9.0, +23.5]). The correlation tests for the secondary outcomes also found a negative correlation trend between the estimates and the coverage of nirsevimab across regions (see Table 3 and Appendix K in S1 Text ).

For each region of the intervention group (N = 4) and the control group, one estimate of the cumulative change during the 2023–2024 season was assessed for each age-group (0–3 and 3–12 months). Those 10 estimates were correlated to the level of coverage of the target-group in each region.

The black points show the observed data. The vertical dotted blue line corresponds to beginning of the intervention period. The solid red line shows the model estimates based on observed data (quasi-Poisson regression modeling) and is associated to its 95%CI in transparent red. The dashed blue line shows the expected values if nirsevimab had not been implemented during the intervention period (quasi-Poisson regression modeling) and is associated to its 95%CI in transparent blue.

The black points show the observed data. The vertical dotted blue line corresponds to beginning of the intervention period. The solid red line shows the model estimates based on observed data (quasi-Poisson regression modeling) and is associated to its 95%CI in transparent red. The dashed blue line shows the expected values if nirsevimab had not been implemented during the intervention period (quasi-Poisson regression modeling) and is associated to its 95%CI in transparent blue.

During the intervention period, the monthly number of bronchiolitis cases presenting to PEDs varied across regions of implementation and across targeted population ranging from +0.6% (95% CI [−9.4, +20.5]) in infants aged 3–12 months in the Paris Region to −61.0% (95% CI [−67.8, −35.2]) in infants aged 0–3 months in the Basque country (see Table 3 and Fig 1 ). Consistently, the reduction in the number of bronchiolitis cases was higher in children aged 0–3 months compared to 3–12 months (see Table 3 and Fig 1 ).

A total of 107,088 infants younger than 1-year old and diagnosed with a bronchiolitis were included during the overall study period (90,619 between January 2018 and September 2023; and 16,469 between October 2023 and March 2024). Among them, 45,904 (42.9%) and 61,184 (57.1%) were included by participating PEDs from the nirsevimab and control group respectively. Bronchiolitis cases aged under 3-months accounted for 36.6% (37,764/103,138) of the cases during the study period. Overall, 33.3% of cases (31,755/95,381) required a hospital admission (see Table 2 ).

Discussion

This multinational quasi-experimental controlled interrupted time-series analysis found a strong correlation between the level of coverage of nirsevimab in the population of infants born in- and out-of-season and the decrease in bronchiolitis cases aged 0–3 and 3–12 months presenting to PEDs across European countries, respectively.

Before the implementation of nirsevimab, modeling [24] studies warned that discrepancies in targeted population or immunization season could highly modify the impact of nirsevimab. However, during this first year of implementation in Europe, neither the targeted populations nor the timing of nirsevimab immunization varied across regions. By contrast, the coverage was highly variable across age categories and regions implementing nirsevimab, and we found it to be correlated to the level of reduction in the monthly number of bronchiolitis cases presenting to PEDs. This correlation was supported by sensitivity analyses. Moreover, the association was also observed between coverage and the reduction in the number of hospitalized bronchiolitis as well as the RSV positivity rate. Thanks to its multinational design, this study provides valuable insights by showing that coverage emerges as a key factor associated to the heterogeneous reduction in bronchiolitis cases after nirsevimab introduction, despite similar implementation guidelines across Europe during the 2023–2024 season.

This study underscores the importance of understanding the multiple factors shaping uptake (local recommendations, pricing, availability, parental acceptability, etc. [29]) in order to translate clinical efficacy into real-world population benefits. First, both France and Spain showed lower coverage in the population of infants born out-of-season compared to those born in-season. Infants born in-season were offered the immunization directly in the maternity ward during RSV season. This setting, providing a direct access to the parents of a new born, especially during the peak season of bronchiolitis, seems favorable compared to the catch-up immunization program.

Regarding the catch-up strategies, in both France and Spain, nirsevimab immunization was free of charge for the patient during the study period. However, the degree of proactive outreach to families to immunize out-of-season infants was very different between France and the different regions of Spain. The different autonomous regions of Spain would actively invite the parents, by text or emails, to have their child immunized, whereas in France, the proposition relied more heavily on the primary care physician’s recommendations. These organizational differences likely contributed to the heterogeneity in coverage observed across regions. In addition, France experienced a shortage of nirsevimab doses for out-of-season births only two weeks after the start of the immunization campaign. This shortage likely reflected higher-than-anticipated demand for nirsevimab. This poor coverage in infants born out-of-season (14%) was reflected in the stable monthly number of bronchiolitis cases aged 3–12 months presenting to Paris Region PEDs in our study (+0.6%, 95% CI [−9,4; +20.5]). Overall, our findings suggest that the real-world impact of nirsevimab may depend as much on implementation choices as on the product’s intrinsic efficacy.

Moreover, beyond its real-world effectiveness on RSV-bronchiolitis across the full spectrum of disease severity [6,10,14], its broader impact on RSV-related infections such as acute otitis media [30,31] has recently been highlighted. Taking this into account alongside the results from our study, prioritizing specific target populations with nirsevimab could influence the impact of nirsevimab on age-related healthcare resource utilization. Specifically, prioritizing infants born in-season would primarily affect the 0–3 months age group, which accounts for the highest RSV burden in hospital settings [32]. Conversely, prioritizing infants born out-of-season would impact most the RSV-related burden in ambulatory care [6,30] and antibiotic use [33]. Overall, this underscores the critical need of defining the clear public health objectives of nirsevimab immunization to maximize its impact.

This study has several limitations. First, this is an observational study, so a definitive causal relationship between the level of coverage of nirsevimab and the reduction in bronchiolitis cases at PEDs cannot be formally established. Second, the identification of cases was based on the ICD-10 codes of the emergency department record and cases were not reviewed manually. However, these codes have been previously evaluated for the surveillance of bronchiolitis [34]. Moreover, the coding and extraction methodology remained unchanged throughout the study period in the different centers; which is a critical point in interrupted time series analysis. Third, cofounding factors not related to nirsevimab outcome (year-to-year variation, late impact of non-pharmaceutical intervention implementation, etc.) could be associated to changes in the consultation patterns and epidemiology of bronchiolitis cases at PEDs. Using a controlled interrupted time-series methodology allowed to assess the evolution of bronchiolitis cases in the nirsevimab group taking into account the evolution of the outcome in the control group. Moreover, no change was observed in the evolution of urinary tract infection cases which served as a control outcome. Fourth, factors related to nirsevimab implementation and influencing RSV epidemiological pattern, such as the emergence of resistant strains to nirsevimab could have influenced our results. However, although some new resistant strains of RSV-B have been recently discovered [35], worldwide RSV surveillance systems have not observed any sign of unusual outbreaks recently [36,37]. Fifth, we hypothesized that the coverage for infants born “in season” would drive the 0–3 months bronchiolitis epidemiology, while the one of “out of season” would drive the 3–12 months bronchiolitis epidemiology. Although these age categories can be discussed, they are used in the real-life management of bronchiolitis every day at PEDs across Europe. Sixth, as no monthly coverage of nirsevimab were available in the regions participating in the study, we accounted for it as a static number (coverage achieved by the end of the immunization season) in the correlation analysis and not in the controlled interrupted time series analysis model. Besides, coverage data may not precisely represent the coverage achieved among the specific population served by each participating PED, as PED-level coverage data were unavailable. This is an inherent limitation of real-world studies conducted at such scale. Seventh, RSV is not the only virus responsible for bronchiolitis. To ensure that we were measuring the association between RSV-bronchiolitis and the anti-RSV strategy, we also assessed the evolution of the RSV positivity rates in the regions from the nirsevimab group and the control group. Eighth, we did not have access to individual-level data on nirsevimab administration and relied on coverage across targeted populations. This was consistent with the objective of our study, which was to evaluate the association between coverage and bronchiolitis reduction at a population-level rather than the individual-level effectiveness or duration of protection. While individual-level analyses are required to characterize protection over time or potential waning of effectiveness, our ecological approach provides a real-world assessment of the association between program implementation and public health benefit.

This multinational interrupted time-series analysis shows heterogeneous impact of nirsevimab on bronchiolitis across European PEDs, identifying coverage as a key factor associated to the reduction. Our results suggest that implementation choices likely influence the population-level benefit through their effect on coverage. These findings highlight that the benefits of nirsevimab depend not only on its intrinsic efficacy, but also on how efficiently it is delivered in real-world settings.

[END]
---
[1] Url: https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1005223

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/