Estimation of the Respiratory Syncytial Virus Hospitalization Burden in Older Adults in the Veneto Region, Italy: A Modeling Study

Abstract

Introduction

Respiratory syncytial virus (RSV) in older adults can cause a variable spectrum of symptoms, ranging from mild manifestations to hospitalization and sometimes adverse outcomes. However, its true epidemiological burden is underestimated due to non-specific symptoms, lack of standardized diagnostic criteria, limited lab confirmation, and inadequate attribution in administrative datasets.

Methods

We conducted a time-series analysis using hospital discharge data from the Veneto Region, Italy, between 2018 and 2024. Respiratory infections (RI) and RSV-related hospitalizations were identified using International Classification of Diseases codes. A generalized additive mixed model (GAMM) was applied to weekly RI admissions, incorporating circulating pathogen data from the RespiVirNet surveillance system. Seasonal patterns and age-stratified risk were modeled using smoothing terms.

Results

Among individuals aged ≥ 65 years, RSV accounted for an estimated 3.0% to 4.6% of RI hospitalizations. Age-specific hospitalization rates attributable to RSV were 26.8, 109.4, and 317.4 per 100,000 person-years in the 65–74, 75–84, and ≥ 85 age groups, respectively. Explicit RSV coding underestimated the true burden by a factor of up to 7.6. Incidence rates and underreporting were highest in post-acute COVID-19 seasons.

Conclusions

RSV-related hospitalizations in older adults are substantially underreported in administrative data. Improved surveillance and prospective clinical studies are needed to validate model estimates and assess diagnostic test performance. Statistical modeling represents a valid approach to estimate the burden of RSV hospitalizations in underdiagnosed populations, such as the elderly, when direct data are lacking.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40121-025-01241-z.

Key Summary Points

Why carry out this study?

RSV infection poses a significant yet under-recognized burden among older adults, particularly those with chronic conditions or immunosuppression.

Current surveillance and hospital coding systems substantially underestimate the true impact of RSV-related hospitalizations in this population.

This study aimed to quantify the hidden burden of RSV hospitalizations in older adults using time-series modeling of regional hospital data from the Veneto Region, Italy.

What was learned from the study?

RSV accounted for 3.0% to 4.6% of respiratory infection–related hospitalizations in individuals aged ≥ 65 years, with underreporting factors up to 7.6 in those aged ≥ 85.

Findings highlight the urgent need for improved diagnostic practices, enhanced RSV surveillance, and expanded adult vaccination strategies to reduce preventable morbidity.

Introduction

Respiratory syncytial virus (RSV) is a viral pathogen that can cause respiratory disease, including bronchiolitis and pneumonia. RSV occurs in infants and adults, with seasonal epidemics typically observed in the colder months.

In adults, particularly older, poly-pathological, and immunocompromised individuals, RSV infection can lead to serious complications, increased morbidity, and higher hospitalization rates [1, 2].

The precise epidemiology of RSV-related disease in adults remains under investigation as more data become available. A meta-analysis estimated that in 2019, RSV caused 5.2 million cases, 470,000 hospitalizations, and 33,000 deaths in adults ≥ 60 years across high-income countries [3]. Recent U.S. surveillance data documented over 16,500 RSV-related hospitalizations, with the highest rates and mortality among those ≥ 75 years [4]. European studies also report a considerable burden: in an Italian pilot study in primary care, more than 21% of acute respiratory infections cases tested RSV-positive, with 3% requiring hospitalization [5], while in Germany, older adults accounted for up to 13% of RSV outpatient visits and 12,800 hospitalizations with 1340 deaths in the 2022–2023 season [6].

Current preventive strategies for older adults include vaccination, with three vaccine types approved by regulatory agencies: adjuvanted recombinant, recombinant, and mRNA vaccines [7–10]. While in some countries RSV adult vaccination has already been approved or officially recommended by national health authorities [11–15], in Italy it is currently endorsed by scientific societies [2, 16]. Based on available evidence of efficacy and safety, the Italian Board of the Calendar of Life recommends RSV vaccination for individuals aged ≥ 75 years and for those aged ≥ 60 years with chronic conditions [17].

Despite the growing awareness of RSV infection over the past two decades, its impact on older adults and other at-risk populations remains insufficiently recognized [18]. Accurate estimates of RSV disease burden are crucial for informing healthcare resource allocation, enhancing prevention strategies, and evaluating the impact of RSV prophylactic interventions, particularly following the recent approval of new vaccines. However, obtaining robust estimates of RSV-associated hospitalizations in older adults has long been recognized as challenging due to underdiagnosis [19–22], driven by non-specific symptoms, lower testing rates compared to children, limited sensitivity of rapid antigen and serological tests in older populations, and underuse of PCR-based diagnostics due to cost constraints [3, 23, 24]. Surveillance limitations and the absence of a standardized clinical case definition for adult RSV further complicate accurate burden assessment [24, 25]. Additionally, hospital records often do not report RSV as the primary cause of hospitalization in patients with comorbidities [24, 26], and many healthcare systems rely on influenza-based surveillance frameworks that may miss atypical or delayed RSV presentations [3].

To address these gaps, statistical modeling approaches have been developed over time to infer RSV-attributable outcomes using indirect methods [6, 23, 25–29]. A common strategy involves correlating respiratory outcomes with RSV circulation proxies while adjusting for confounders such as co-circulating pathogens and seasonal trends using generalized linear models [24].

We aim to address the gap in knowledge of RSV epidemiology in older adults in the Veneto Region, providing evidence that may support healthcare planning and future evaluations of intervention impact. Specifically, our study aims to indirectly estimate the burden of RSV-associated hospitalizations among adults aged ≥ 65 years in the Veneto Region, Italy, using a time series approach that combines hospital discharge data with virological trends from the RespiVirNet surveillance system to account for underreporting. The analysis covers the period from 2018 to spring 2024, allowing comparisons between pre- and post-COVID-19 emergency seasons.

Methods

Study Design and Population

We conducted a retrospective analysis of regional hospitalization and emergency department (ED) admissions databases to estimate the incidence of RSV-attributable hospitalizations in the Veneto Region, Italy. In 2024, the Veneto Region had approximately 4.8 million inhabitants, with an average age of 46.9 years, placing it 13th among Italy’s 19 regions and two autonomous provinces in terms of population aging [30–34]. Notably, 24.5% of the population is aged 65 years and older, and the region has a dependency ratio of 57.5.

Healthcare in the Veneto Region is provided through Italy’s National Health Service, which is funded by general taxation [32, 33, 35]. Service delivery is decentralized, with Local Health Authorities responsible for care coordination and provision.

Data Sources

In this study, we identified all respiratory-related hospital and ED admissions in the Veneto Region, using specific International Classification of Diseases (ICD) codes, as defined in similar model-based studies by the RESCEU (Respiratory Syncytial Virus Consortium in Europe) and PROMISE (Preparing for RSV Immunisation and Surveillance in Europe) consortia [26–28]. We retrieved cases from hospital discharge records (HDRs) and ED regional administrative databases recorded between 2015 and 2024. As previously reported for the same data sources [32], diagnoses in the Veneto Region are encoded using the ICD 9th revision system (ICD-9-CM [36]). Details on the outcome definitions and adopted conversions from the ICD 10th revision (ICD-10-CM [37]) to the 9th revision system are provided in Supplementary Table S1.

A respiratory admission was defined as any HDR or ED record containing at least one respiratory infection-specific ICD-9-CM code. In particular, RSV-coded admissions were further classified using ICD-9-CM codes: 079.6 (RSV infection), 480.1 (pneumonia due to RSV), and 466.11 (acute bronchiolitis due to RSV), similar to [23, 38].

We included admissions for individuals of all ages, from all inpatient facilities, both public and private, within the Veneto Region, regardless of admission type or hospitalization regimen. Records of individuals not residing in the Veneto Region were excluded. To improve the accuracy of our estimates, we also excluded repeated hospitalizations for the same individual occurring within 30 days of a prior discharge, reducing potential double-counting from overlapping treatment periods.

To enhance the estimation of the RSV-associated burden, virological data were retrieved to enable statistical modeling of the impact of virus circulation on hospital admission rates.

In the Veneto Region, virological surveillance is conducted under the protocol of the national integrated surveillance system of respiratory viruses, known as RespiVirNet (formerly InfluNet) [39, 40], coordinated by the Istituto Superiore di Sanità (ISS) with support from the Italian Ministry of Health. RespiVirNet operates through a community-based network of sentinel general practitioners, pediatricians, and regional reference laboratories, using an influenza-like illness (ILI) case definition [40].

RespiVirNet system coverage and data volumes have improved over time, especially following the SARS-CoV-2 pandemic. In addition to influenza viruses, since the 2019–2020 influenza season, the Veneto Region has also reported RSV cases (at the national level, RSV reporting began in 2022). SARS-CoV-2 has been included since 2020, and from 2022, additional respiratory viruses have been added, including rhinoviruses, non-human coronaviruses, metapneumoviruses, adenoviruses, parainfluenza viruses, and bocaviruses.

For this study, we used RespiVirNet surveillance data collected during the influenza seasons from 2018–2019 through 2023–2024 in the Veneto Region.

Data were aggregated into age-specific weekly counts of hospital discharges, ED admissions, and positive tests, based on the following age groups: 0–11 months, 1–2 years, 3–4 years, 5–17 years, 18–64 years, 65–74 years, 75–84 years, and 85 + years. The intersection of available time series allowed us to cover the period from 2018 to 2024.

Modeling

To estimate the true burden of respiratory hospitalizations among adults, we adopted a modeling approach similar to that used in previous studies [23–28]. A generalized additive mixed model (GAMM) was employed to estimate the weekly number of respiratory hospital discharges in association with major circulating respiratory pathogens, Influenza A, Influenza B, RSV, and, for the pandemic period, SARS-CoV-2, represented by weekly counts of positive cases. In addition to pathogen covariates, a cyclic penalized cubic spline was applied to calendar week to capture seasonal trends. Random effects were included for age group and influenza season using smooth terms to allow for differences in baseline hospitalization rates across age classes and seasonal variations. These were treated as parametric random effect terms penalized using a ridge penalty, contributing to model flexibility while avoiding overfitting. To account for delayed or early effects of viral circulation, lagged term predictors were also included.

Separate models were fitted for distinct periods to allow flexibility in capturing changing epidemiological patterns before, during, and after the COVID-19 emergency, as well as differences in data reporting within the RespVirNet surveillance system. The periods were defined as follows: pre-pandemic (week 46 of 2018 to week 8 of 2020); COVID-19 acute phase with strong public health restrictions (week 9 of 2020 to week 14 of 2022); and post-acute phase, characterized by the relaxation of control measures and the transition into the post-pandemic period (week 15 of 2022 to week 26 of 2024). This classification was intended to capture the impact of the COVID-19 emergency on surveillance systems, including disruptions in data reporting, differences in testing practices, hospitalizations, and changes in viral circulation driven by the implementation and subsequent relaxation of non-pharmaceutical interventions. It was informed by regional and national epidemiological data and regulatory measures [41–43].

For all age groups and periods, the models hence followed the form of

$$\text{log}\left(Y_{a,t}\right)=s_a\left(t\right)+{\beta }_{a,0}+{\beta }_{0,z}+\sum _{i=0}^4{\beta }_{a,1_i}{\text{Influenza A}}_{a,t-i}+\sum _{i=0}^4{\beta }_{a,2_i}{\text{InfluenzaB}}_{a,t-i}+\sum _{i=0}^4{\beta }_{a,3_i}{\text{RSV}}_{a,t-i}+\sum _{i=0}^4{\beta }_{a,4_i}{\text{SARS - CoV - 2}}_{a,t-i}$$

where $Y_{a,t}$ represents the weekly number of respiratory hospitalizations for age group $a$ in week $t$, $s_a(t)$ represents the penalized spline function in age group $a$, ${\beta }_{a,0}$ and ${\beta }_{0,z}$ are the random effects for age group $a$ and influenza season $z$, and ${\beta }_{{a,p}_i}$ are the pathogen $p$—and age-specific $a$ regression coefficients for weekly positive cases in week $t$ or lagged values with $i$ weeks of delay. The model assumes that the observed cases follow a quasi-Poisson distribution,

$${\gamma }_{a,t}\sim \text{quasi-Poisson}(Y_{a,t})$$

to account for potential overdispersion in the count data.

Since specific case counts for RSV were not available for the 2018–2019 season, the number of RSV-coded ED admissions was used as a proxy measure to impute 2018–2019 RSV activity. The model was implemented using the “mgcv” package in R [44].

Model and variable selection, particularly for lagged terms, was guided by performance comparisons. Model performance was evaluated based on the lowest root mean squared error (RMSE), visual inspection of residuals, the statistical significance of smooth and linear terms, and the adjusted R².

Finally, for each age group and period, the weekly number of RSV-attributable respiratory hospitalizations was estimated by calculating the difference between the predicted values from the full model and those from a reduced model in which all RSV-related terms were set to zero, as in [23, 27]. Negative differences were set to zero.

Statistical Analysis of RSV-Related Outcomes

We calculated the annual number of respiratory and RSV-related admissions identified by ICD-9-CM codes by summing weekly cases within each influenza season and age group. The annual number of RSV-predicted cases was obtained analogously.

Age-specific annual incidence rates per 100,000 individuals were calculated by dividing the annual number of cases by the population at risk, expressed in person-years. The population at risk was defined as the number of age-specific populations of the Veneto Region (source [30]), multiplied by the fraction of the year corresponding to the duration of the influenza season (approximately late October to mid-April, as defined by the RespiVirNet protocol [40]).

We also calculated the percentage of respiratory admissions attributable to RSV by dividing the total number of RSV admissions (ICD-coded or predicted) by the total number of respiratory-related admissions within each age group.

Finally, a multiplicative correction factor for RSV-coded admissions was computed by dividing the number of predicted RSV admissions by the number of RSV-coded admissions for each age group and season. A ratio greater than 1 indicates potential underreporting or underdiagnosis of RSV.

Although the study focuses on adults aged ≥ 65 years, pediatric age groups (≤ 2 years) were included in the analysis, using their multiplicative factors as a proxy for model accuracy, since underestimation is less pronounced in this population [27].

Estimates of annual incidence rates over the entire study period, 2018–2024, were calculated as weighted averages, using person-years at risk for each season as weights. Overall RSV-attributable proportions of respiratory admissions and correction factors were computed by dividing the total number of cases over the entire period.

The acute COVID-19 season, 2020–2021, was excluded from the outcome analysis due to confounding factors previously described, including uncertainties in viral circulation, disruptions in surveillance, and the impact of public health containment measures. Sensitivity analyses were additionally conducted by stratifying the outcomes by periods, pre-pandemic seasons (2018–2020) versus post-acute COVID-19 seasons (2021–2024), and by excluding the 2021–2022 season, during which public health restrictions had been relaxed but the state of emergency had not yet formally ended. Although this classification does not exactly follow the official WHO SARS-CoV-2 emergency timeline [45], it reflects the regional epidemiological context and the gradual resumption of routine healthcare and surveillance practices [41–43].

Joinpoint regression analyses [46] were conducted to assess trends in health outcomes. Results are reported as average annual percentage change (AAPC), with corresponding 95% confidence intervals (CI). A p value < 0.05 was considered statistically significant.

All data processing, statistical analyses, and visualizations were performed using Python version 3.8.18 (Python Software Foundation, Wilmington, DE, USA) and R version 4.2.2 (R Core Team, Vienna, Austria).

Ethics Statement

This study was conducted in accordance with relevant ethical guidelines. Aggregated data were obtained from databases of the Veneto Region, for which official permission to access and use the data was granted by the Regional Health Authority, and the disclosure and utilization of such records for educational and scientific purposes do not necessitate approval from ethical committees. On January 24, 2023, the Veneto Region implemented the Code of Conduct for the use of health data for educational and scientific publication purposes (Official Bulletin of the Region, “Bollettino Ufficiale della Regione” no. 10), as established by the European Committee (European Regulation 2016/679). This implementation previously received approval from the Italian Personal Data Protection Authority on January 14, 2021.

Adhering to the current Italian privacy legislation, the publication and utilization of health data, along with the processing methods, must occur exclusively in aggregate form, without any reference to patients’ personal information. Prior to providing access to the authors, all personal data that could potentially lead to identification was substituted with anonymous codes, in accordance with current privacy regulations (Legislative Decree no. 196 of June 30, 2003).

Results

Respiratory Infection-Coded Hospitalizations

A total of 113,748 hospital discharges for respiratory infections were reported in the Veneto Region between the 2018–2019 and 2023–2024 influenza seasons. Individuals aged ≥ 65 years accounted for 72.52% (82,500) of all hospitalizations. The trend of respiratory infection-related hospitalizations showed strong seasonality, with peaks typically occurring in the winter months of December and January (Fig. 1 and Supplementary Material Figure S1). Exceptions were observed in 2020 and 2021, during the COVID-19 acute phase, in which unusual or multiple peaks were recorded around March or April. The COVID-19 emergency seasons were also characterized by an increase in respiratory-related hospitalizations among adults over 18 and a marked reduction in pediatric admissions (Fig. 1; Supplementary Material Figure S1 and Table S2).

Fig. 1.

Weekly number of hospitalizations for respiratory infections and respiratory syncytial virus (RSV)-related hospital discharges in the Veneto Region from 2018 to 2024, stratified by age group: 65–74 years (A), 75–84 years (B), and ≥ 85 years (C). Dots represent observed (coded) cases, while the solid line represents model predictions (left y-axis). Annual incidence rates per 100,000 person-years are displayed as dashed lines (right y-axis). Panels A and B include an inset (top right) with a zoomed view of the RSV traces. The shaded area approximately corresponds to the COVID-19 acute phase, during which strong public health restrictions were in place in Italy

Excluding the acute COVID-19 season, 2020–2021, a decreasing trend in respiratory infection hospitalizations was observed, with negative AAPCs, although not statistically significant (Supplementary Material Table S2). The annual incidence rate decreased from 937.71 to 860.51 per 100,000 person-years among individuals aged 65–74, from 2605.70 to 2336.07 among those aged 75–84, and from 7627.65 to 6507.34 among those aged 85 and over, when comparing pre-pandemic (2018–2020) with post-acute COVID-19 seasons (2021–2024) (see Table 1).

Table 1.

Absolute number and incidence rate (per 100,000 person-years) of hospitalizations for respiratory infections (RI), with coded versus predicted respiratory syncytial virus (RSV)-attributable cases in older adults (≥ 65), by age group

Seasons	Age (years)	RI hospitalizations		RSV-coded hospitalizations	RSV-predicted hospitalizations			Multiplicative factor
		n	Rate × 100,000	n	n	Rate × 100,000	RSV attributable (%)
Pre-pandemic seasons (2018–2020)	0	2370	7855.79	1146	964	3194.81	40.67	0.84
	1–2	916	1433.57	134	34	52.50	3.66	0.25
	65–74	4292	937.71	19	31	6.68	0.71	1.61
	75–84	8800	2605.70	23	115	34.13	1.31	5.01
	85 +	10,931	7627.65	16	118	82.46	1.08	7.39
Post-acute COVID-19 seasons (2021–2024)	0	3514	7074.95	2172	1855	3734.08	52.78	0.85
	1–2	1303	1261.30	345	202	195.70	15.52	0.59
	65–74	7403	860.51	104	322	37.48	4.36	3.10
	75–84	14,870	2336.07	138	951	149.38	6.39	6.89
	85 +	18,055	6507.34	160	1217	438.76	6.74	7.61
All seasons (excluding 2020–2021)	0	5884	7370.02	3318	2818	3530.30	47.90	0.85
	1–2	2219	1327.13	479	236	140.97	10.62	0.49
	65–74	11,695	887.32	123	353	26.78	3.02	2.87
	75–84	23,670	2429.53	161	1066	109.43	4.50	6.62
	85 +	28,986	6888.91	176	1336	317.41	4.61	7.59

Data for infants (0 years) and young children (1–2 years) are also included to allow comparison of incidence rates and underestimation factors across age groups

RI respiratory infections, RSV respiratory syncytial virus, n total number of hospitalizations over the considered seasons

RSV-Coded Hospitalizations

A total of 4572 RSV-specific hospitalizations were recorded in the Veneto Region based solely on RSV-specific ICD coding. The majority of cases occurred in infants under 1 year of age (72.62%), followed by children aged 1–2 years (10.48%) and adults aged 65 years and older (10.19%), following a U-shaped age distribution. Notably, only eight cases were recorded during the 2020–2021 season (Supplementary Material Table S2). Excluding this period, the annual incidence among individuals aged ≥ 65 ranged from 3.92 to 94.82 cases per 100,000 person-years, representing only about 0.05–2.48% of all hospitalizations due to respiratory infections (Supplementary Material Table S2).

In contrast to the overall trend observed for all respiratory infections, RSV-coded hospitalizations showed an increasing trend across all age groups, with statistically significant positive AAPCs for the 1–2 year, 65–74 year, and 75–84 year age groups (Fig. 1; Supplementary Material Figure S1 and Table S2). Specifically, the incidence increased from 4.38 to 9.95 per 100,000 person-years between the 2018–2019 and 2023–2024 seasons in individuals aged 65–74 years (AAPC: 30.89; 95% CI 0.178, 70.426; p = 0.048), from 6.57 to 14.20 in those aged 75–84 years (AAPC: 31.60; 95% CI 0.848, 70.218; p = 0.041), and from 4.23 to 50.00 in individuals aged ≥ 85 years (AAPC: 68.08; 95% CI − 4.868, 198.293; p = 0.078). For all age groups, the highest incidence of RSV-coded hospitalizations was observed during the 2022–2023 season, with rates of 20.55, 41.14, and 94.83 per 100,000 person-years in the 65–74, 75–84, and 85 + age groups, respectively.

RSV-Predicted Hospitalizations and Underreporting

Between the 2018–2024 influenza seasons, excluding 2020–2021, a total of 6864 hospitalizations were attributed to RSV in the Veneto Region according to the predictive model, representing a 50.13% increase compared to RSV-coded cases. The predicted RSV cases confirmed the typical U-shaped age distribution, with the majority occurring in infants under one year (41.07%) and adults aged 65 years and older (40.13%).

Figure 1 illustrates that the number of predicted RSV hospitalizations (red solid line) generally differs from the observed RSV-coded hospitalizations (red dots) among adults aged 65 and older. Moreover, the close agreement between observed and predicted respiratory infection-related hospitalizations (blue dots and blue solid lines, respectively) across all age groups supports the good overall fit of the model (Fig. 1 and Supplementary Material Figure S1).

For the study period, excluding the 2020–2021 season, the model estimated annual RSV-related incidence rates of 26.78, 109.43, and 317.41 per 100,000 person-years in the 65–74, 75–84, and 85 + age groups, respectively (Table 1). According to these estimates, RSV would be attributable to 3.02%, 4.50%, and 4.61% of all hospitalizations for respiratory infections in these age groups.

When calculating the multiplicative factor as the ratio of RSV-predicted to RSV-coded hospitalizations, values greater than one were observed across all adult age classes, indicating an underestimation of the true RSV burden (Table 1 and Supplementary Material Table S2). The multiplicative factor generally increased with age. Based on our statistical approach, RSV-coded hospitalizations underestimated the actual number of RSV-attributable respiratory hospitalizations by 2.87, 6.62, and 7.59 times in the 65–74, 75–84, and 85 + age groups, respectively, with an overall underestimation factor of 5.99 among individuals aged 65 years and older.

In contrast, in children ≤ 2 years, the multiplicative factor was slightly below one, indicating that predicted and observed values are closely aligned (Table 1 and Supplementary Material Figure S1). This suggests that RSV-specific ICD codes serve as a good proxy for identifying RSV hospitalizations in early childhood.

Finally, a comparison of predicted incidences in pre-pandemic and post-acute COVID-19 seasons reveals a general increase in RSV-related hospitalizations in both pediatric and adult populations. Among adults aged 65–74, the incidence rose from 6.68 to 37.48; in those aged 75–84, from 34.13 to 149.38; and in individuals aged 85 and older, from 82.46 to 438.76 (Table 1). Joinpoint trend analysis showed positive AAPCs for all age groups and values exceeding 30 in the over-65 population, although significance was reached only for the 1–2 years and 75–84 years age groups (Supplementary Material Table S2). Furthermore, as observed for RSV-coded hospitalizations, the predicted hospitalizations peaked during the 2022–23 season in both children under one year of age and adults over 65, reaching rates of 252.40 and 852.72 per 100,000 person-years in the 75–84 and 85 + age groups, respectively.

In older adults, the (underreporting) multiplicative factors also increased in the post-acute COVID-19 phase, particularly in the 65–74 age group (from 1.61 to 3.10) and in the 75–84 age group (from 5.01 to 6.89, Table 1). Considering the overall decreasing trend in hospitalizations for respiratory infections, these predictions yield to growing proportions of cases attributable to RSV, with the percentage in individuals aged 65 and older increasing from 1.10% to 6.18% between the pre-pandemic and post-acute COVID-19 periods.

Discussion

In this study, we estimated the burden of RSV-associated hospitalizations among older adults in the Veneto Region over six influenza seasons (2018–2024) using a time series modeling approach that integrates ICD-coded hospital discharge data with virological surveillance. Our results indicate that routine RSV-specific coding substantially underestimates the true burden of RSV in adults aged ≥ 65 years. According to our model, the actual annual RSV-related hospitalization rates ranged from 26.78 to 317.41 per 100,000 person-years, representing 3.02% to 4.61% of all hospitalizations for respiratory infections in this age group. Based on our statistical approach, the predicted number of RSV-attributable hospitalizations in this population was 5.99 times higher than the number identified through diagnostic coding alone, with increasing underestimation observed in older age strata: 2.87-fold in those aged 65–74, 6.62-fold in those aged 75–84, and 7.59-fold in those aged 85 and older.

These findings align with international evidence highlighting the significant and largely underrecognized impact of RSV on older adults [24]. A 2022 similar modeling study by the RESCEU group estimated that RSV accounts for more than 3–7.2% of all respiratory admissions in adults aged ≥ 65 years across six Northern European countries [28]. Osei-Yeboah et al. extrapolated these estimates to 28 European countries using nearest-neighbor matching techniques, reporting rates between 0.66 and 2.99 per 1000 person-years [26]. In an updated analysis incorporating SARS-CoV-2 activity, Johannesen et al. found RSV-predicted-to-coded hospitalization ratios of 1.9–33 for adults aged ≥ 65 years, generally increasing with age [27]. The authors also applied their time-series approach to data from a hospital-based surveillance network in Valencia, Spain, which identified cases based on ILI criteria rather than ICD codes and tested all hospitalized patients for RSV. In this setting, the degree of underestimation was lower (ratios 2.1–2.7), likely due to more systematic viral testing in hospitalized cases.

In Germany, Scholz et al. reported a sevenfold underdetection in hospitalizations and a 15-fold increase in in-hospital mortality among adults aged ≥ 60 years [6].

In Italy, Méroc et al. estimated the RSV-attributable incidence of various cardiorespiratory outcomes using a quasi-Poisson regression model based solely on hospitalization data, defining viral activity proxies for RSV and influenza as specifically coded hospitalizations in specific populations with more frequent testing [23]. Their results indicate a substantial burden beyond respiratory complications, with mortality 2–3 times higher for cardiorespiratory outcomes compared to respiratory disease alone, with the highest rates observed in individuals aged ≥ 75 years (59–85 deaths per 100,000 person-years). Incidence rates based on RSV-specific ICD codes alone were 405–1729-times-lower than predicted estimates accounting for untested cases, reaching 1064–1527 cases per 100,000 person-years in adults aged ≥ 75 years.

Despite variability—likely due to country- and age-specific differences in coding practices, viral circulation, and surveillance systems—these findings consistently underscore the challenges in clinically identifying and reporting RSV in older adults. In contrast, RSV ICD-coded hospitalizations in children matched more closely model predictions, with a multiplicative factor generally slightly below or near one, similar to [24, 27]. This confirms the accuracy of ICD codes in identifying RSV hospitalizations in early childhood and suggests minimal underreporting, underdiagnosis, or misclassification in this age group.

Available evidence also agrees on the substantial burden observed in adults, suggesting that vaccination could have a significant positive impact, and highlighting the importance of real-world data to assess the population-level effects of newly approved vaccines and to optimize future immunization strategies.

Epidemiological modeling approaches have demonstrated their value in understanding the true disease burden in adults, showing satisfying consistency with prospective studies [24]. However, results heterogeneity depending on the case definitions used for the respiratory outcome variables, on the adjusting covariates, or on their time lags highlights the need for further prospective cohort studies with active case testing to validate indirect statistical estimates with real data [24].

From a more technical perspective, regression modeling methods are still the most commonly used approach for estimating RSV-associated excess disease burden, which may require longer and more stable time series for both outcomes and covariates [24]. This aspect may be relevant when interpreting results based on variables representing phenomena with underlying changes, such as healthcare activity or surveillance, as observed during the COVID-19 pandemic. In the Veneto Region, we observed a decreasing trend in hospitalizations for respiratory infections. In contrast, RSV-coded hospitalizations showed an increasing trend across all age groups, excluding the 2020–2021 season, consistent with a previous study on the same population [38]. The predicted underreporting multiplicative factors also increased in the post-acute COVID-19 phase, particularly in the 65–74 age group (from 1.61 to 3.10) and in the 75–84 group (from 5.01 to 6.89). These trends have also been observed in other countries, such as Germany and Finland [6, 27, 47], though they are not consistent across all settings [4, 27]. In our region, post-pandemic estimates are expected to have less uncertainty, as they are based on a larger number of surveillance observations. Further longitudinal research is required to provide additional data and estimates on RSV circulation and burden in the post-pandemic era. Such evidence would determine whether the observed differences are due to actual changes in pathogen spread or infection severity, or whether they are attributable to more complex dynamics, including differences in public health policies, enhanced surveillance, or improved diagnostic practices.

This study has both strengths and limitations. First, it is a population-based analysis covering an entire administrative region over six consecutive influenza seasons (2018–2024), allowing comparisons between pre- and post-pandemic periods. It relies solely on real-world, age-specific data from structured regional administrative databases, with cases identified using ICD codes, and from the regional virological surveillance system, ensuring reproducibility. Our methods could serve as a framework for other Italian regions or settings with similar administrative data in the absence of direct evidence.

Nonetheless, the results discussed rely on indirect methods, and there are inherent modeling limitations to consider. The analysis focused on hospitalizations and did not include more severe outcomes, such as ICU admissions or mortality, which could provide a more complete picture of disease burden. The modeling approach also did not explicitly account for co-infections or incorporate aggregated clinical and socio-demographic variables, including vaccination coverage. Geographical heterogeneity within the region could not be assessed, as surveillance data were not available at the provincial or municipal level, limiting the ability to adjust for healthcare system differences or environmental factors. In addition, although random effects for age and influenza season were included, the flexibility of the GAMM approach may not fully capture age-specific differences in the relationship between virus circulation and hospitalizations, possibly leading to some underestimation of cases. Finally, our time-series approach did not explicitly incorporate population structure, such as age-stratified contact rates or mobility patterns, which may play a significant role in RSV transmission and hospitalization dynamics.

Conclusions

Our study confirms that RSV contributes significantly to respiratory-related hospitalizations among adults, particularly in older populations, with considerable underreporting in routine hospital records. Using a time-series modeling approach, we estimated an average underdetection factor of approximately six for individuals aged ≥ 65 years during the 2018–2024 period, with larger values observed in the post-acute COVID-19 phase. These region-specific estimates may serve as correction factors for future analyses, including pharmacoeconomic evaluations and vaccine effectiveness studies.

Statistical modeling offered a flexible and reproducible method to quantify the true burden of RSV hospitalizations in the Veneto Region in underdiagnosed populations such as the elderly in the absence of direct evidence. However, it remains necessary to strengthen RSV surveillance systems and conduct prospective clinical studies to validate data obtained through indirect modeling approaches. Robust real-world burden estimates are essential to guide health policy decisions in the post-pandemic era.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Conceptualization, Claudia Cozzolino and Vincenzo Baldo; data curation, Laura Salmaso, Mario Saia, Davide Gentili, Francesca Russo, and Michele Tonon; formal analysis, Claudia Cozzolino; funding acquisition, Vincenzo Baldo; investigation, Claudia Cozzolino, Andrea Cozza and Vincenzo Baldo; methodology, Claudia Cozzolino, Andrea Cozza, Silvia Cocchio and Vincenzo Baldo; project administration, Vincenzo Baldo; resources, Mario Saia, Davide Gentili, Michele Tonon, Francesca Russo, and Vincenzo Baldo; software, Claudia Cozzolino; supervision, Mario Saia, Davide Gentili, Michele Tonon, Francesca Russo, and Vincenzo Baldo; validation, Andrea Cozza, Silvia Cocchio and Vincenzo Baldo; visualization, Claudia Cozzolino; writing—original draft, Claudia Cozzolino and Andrea Cozza; writing—review and editing, Silvia Cocchio, Davide Gentili, Michele Tonon, Francesca Russo, and Vincenzo Baldo. All authors have read and agreed to the published version of the manuscript.

Funding

This research received unrestricted funding from Moderna Italy (Rome, Italy). The Rapid Service Fee was also funded by Moderna Italy. No other funding from industry or public sources was received, and the authors declare no conflicts of interest.

Data Availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Code Availability

The codes that support the findings of this study are available upon request addressed to the corresponding author.

Declarations

Conflict of Interest

The authors declare no competing interests. Claudia Cozzolino, Andrea Cozza, Laura Salmaso, Mario Saia, Davide Gentili, Michele Tonon, Francesca Russo, Silvia Cocchio, Vincenzo Baldo.

Ethical Approval

This study was conducted in accordance with relevant ethical guidelines. Aggregated data were obtained from databases of the Veneto Region, for which official permission to access and use the data was granted by the Regional Health Authority, and the disclosure and utilization of such records for educational and scientific purposes do not necessitate approval from ethical committees. On January 24, 2023, the Veneto Region implemented the Code of Conduct for the use of health data for educational and scientific publication purposes (Official Bulletin of the Region, “Bollettino Ufficiale della Regione” no. 10), as established by the European Committee (European Regulation 2016/679). This implementation previously received approval from the Italian Personal Data Protection Authority on January 14, 2021. Adhering to the current Italian privacy legislation, the publication and utilization of health data, along with the processing methods, must occur exclusively in aggregate form, without any reference to patients’ personal information. Prior to providing access to the authors, all personal data that could potentially lead to identification was substituted with anonymous codes, in accordance with current privacy regulations (Legislative Decree no. 196 of June 30, 2003).