Respiratory Syncytial Virus–Associated Hospitalization in Adults With Comorbidities in 2 European Countries: A Modeling Study

Abstract

Background

Individuals with comorbidities are at increased risk of severe respiratory syncytial virus (RSV) infection. We estimated RSV-associated respiratory hospitalization among adults aged ≥45 years with comorbidities in Denmark and Scotland.

Methods

By analyzing national hospital and virologic data, we estimated annual RSV-associated hospitalizations by 7 selected comorbidities and ages between 2010 and 2018. We estimated rate ratios of RSV-associated hospitalization for adults with comorbidity than the overall population.

Results

In Denmark, annual RSV–associated hospitalization rates per 1000 adults ranged from 3.1 for asthma to 19.4 for chronic kidney disease (CKD). In Scotland, rates ranged from 2.4 for chronic liver disease to 9.0 for chronic obstructive pulmonary disease (COPD). In both countries, we found a 2- to 4-fold increased risk of RSV hospitalization for adults with COPD, ischemic heart disease, stroke, and diabetes; a 1.5- to 3-fold increased risk for asthma; and a 3- to 7-fold increased risk for CKD. RSV hospitalization rates among adults aged 45 to 64 years with COPD, asthma, ischemic heart disease, or CKD were higher than the overall population.

Conclusions

This study provides important evidence for identifying risk groups and assisting health authorities in RSV vaccination policy making.

Respiratory syncytial virus (RSV) is a major cause of respiratory tract infections (RTIs) and leads to about 245 000 and 158 000 hospitalizations annually in young children and adults aged ≥18 years, respectively, in EU countries, with a disproportionate burden occurring in infants and adults aged ≥65 years [1, 2]. Besides older age, comorbidities are a key risk factor for RSV hospitalization in the adult population. Previous epidemiologic studies show that comorbidities such as chronic respiratory disease, chronic heart disease, and diabetes are prevalent among adult patients hospitalized with RSV-associated RTI [3–6]. Moreover, chronic kidney disease (CKD) and cardiovascular disease are linked with the risk of RSV severe illnesses and mortality [7–9]. A systematic review estimated a high incidence of RSV-associated RTI and a high risk of mortality from hospitalized RSV among adults with comorbidities (eg, cystic fibrosis, congestive heart failure, chronic obstructive pulmonary disease [COPD], or immunosuppression) [10]. However, estimates of RSV-associated RTI hospitalization rates among adults with comorbidity are sparse in terms of geographic diversity [10, 11] and the type of comorbidities being investigated [11], and none are available in the European region. Two RSV vaccines, RSVPreF3-AS01E and RSVpreF, have been approved for use in older adults by the European Medicines Agency [12, 13]. In view of the rapid progress in the development and evaluation of RSV vaccines and prophylaxis products, estimates of RSV hospitalization rates in adults with comorbidity could provide timely evidence for recommendations on RSV immunization among the adult population on the horizon.

Using routinely collected data from national hospital registries and virologic surveillance, we aimed to estimate the number, rates, and risk of RTI hospitalizations associated with RSV in the pre–COVID-19 era among Danish and Scottish adults aged ≥45 years who had at least 1 of 7 comorbidities, as specified in case definitions. The comorbidities were selected by influenza vaccination recommendations and data availability, and we used a regression modeling approach [14]. Regression models are widely used to estimate RSV disease burden, especially among adults. Specifically, they account for (1) underascertainment of RSV diseases, due to the lack of systematic RSV testing [15] and poor sensitivity of RSV-specific ICD-10 codes in routine clinical care practice [16, 17]; (2) the imperfect sensitivity of viral diagnostic tests [18]; and (3) the potentially poor detection of RSV in late-stage disease samples.

METHODS

Study Design and Data Sources

The study design and data sources have been described previously [19]. Briefly, we conducted a retrospective analysis using national hospital registries and virologic surveillance data in Denmark (2010–2018) and Scotland (2010–2016). A season was from week 40 of 1 year to week 39 of the following year.

Case Definitions

As done previously [19, 20], we defined RTI hospitalizations based on ICD-10 diagnosis codes (Supplementary Table 1). RTI admission was defined as an admission with any mention of RTI in the diagnosis codes. We included 7 comorbidities according to Scotland recommendations on high-risk conditions for influenza vaccination and data availability: COPD, asthma, ischemic heart disease (IHD), stroke, diabetes, CKD, and chronic liver disease (CLD) [21]. The comorbidities were identified by ICD-10 diagnosis codes according to previous disease burden study (Supplementary Table 2) [22]. We searched all the diagnostic fields to identify individuals who were hospitalized due to and with the diseases and were recorded on any occasions during health care utilization within 5 years before the RTI hospital episode or at the episode.

Scottish Burden of Disease group estimates of the prevalence of the 7 chronic medical conditions in 2014 were used to derive hospitalization rates in Scotland [22]. To derive hospitalization rates in Denmark, we used the prevalence of IHD, stroke, diabetes, and CKD in Danish adults in 2015 and the prevalence of COPD and asthma in 2014 obtained from Danish disease burden reports and registers [23, 24]. Due to small case counts, Danish adults with CLD were excluded from the study.

Statistical Analyses

Model Overview

Data were accessed and analyzed separately by partners in Denmark and Scotland but were based on the same analytic approach. A multiple linear regression model was used to estimate the number of RTI hospitalizations associated with RSV (“RSV-associated RTI”) in adults aged ≥45 years with chronic medical conditions, similar to previous analyses [19, 25, 26]. Overall, the model included a natural cubic spline function for weeks during the study period to model the long-term trend and seasonal pattern of RTI hospitalizations and the number of RSV-positive test results; the number of influenza-positive test results were also included to account for the confounding effect of cocirculating influenza. In Scotland, we considered an interaction term of influenza-positive test results and season, as there was a greater number of influenza-positive test results in the 2010–2011 season as compared with other seasons, which may reflect changes in testing practices over the study period. We separately modeled chronic medical conditions in the 2 countries; for each condition, we also conducted subgroup analyses by age (45–54, 55–64, 65–74, 75–84, and ≥85 years), where appropriate. For each comorbidity and age group, we tested for the optimal lag/lead combination on RSV and influenza among a lag/lead of 0 to 3 weeks. The goodness of fit was assessed by adjusted R² and the Akaike information criterion. Details of the model structure are given in the supplementary file.

We estimated the annual number of RSV-associated RTI hospital admissions based on model coefficients for RSV and the number of RSV-positive test results. The 95% CIs were estimated with a 52-week-block bootstrap with 1000 replicates. We estimated rate ratios (RRs) of RSV-associated RTI hospitalization between the comorbid population and the overall population (with or without the comorbidity) of the same age band. Estimated 95% uncertainty ranges (URs) of RRs were based on 1000 samples generated from a log-normal distribution of estimated hospitalization rates, with the 2.5th percentile of the samples as the lower bound and the 97.5th percentile as the upper bound, as previously done [27].

Sensitivity Analyses

We conducted the following sensitivity analyses among adults aged ≥45 years to assess the robustness of RSV burden estimates: (1) assuming zero lags for RSV and influenza predictors; (2) removing the influenza predictor from the models of which the coefficient was negative; (3) using the Poisson regression model; and (4) adding time series data of rhinovirus-positive test results to the models to account for a potential confounding effect for Scottish population, based on empirical evidence that rhinovirus is identified in a proportion of RTI hospitalizations among adults, especially those with comorbidities [28, 29].

Ethical Statement

Data access approvals were obtained in both countries, and analyses were conducted in each country in a secured environment.

RESULTS

Average Annual Number and Proportion of RSV-Associated RTI Hospitalizations in Adults Aged ≥45 Years With Comorbidities

We estimated that, in Danish adults aged ≥45 years with comorbidities, the average annual number of RSV-associated RTI hospitalizations ranged from 165 (95% CI, 160–218) for CKD to 1876 (95% CI, 1613–2178) for COPD (Supplementary Table 3). RSV-associated RTI accounted for 5.8% (95% CI, 4.5–7.3) of all RTI hospitalizations among adults with stroke and 16.8% (95% CI, 14.2–19.4) with asthma (Supplementary Table 4).

In Scottish adults aged ≥45 years with comorbidities, the average annual number of RSV-associated RTI hospitalizations ranged from 64 (95% CI, 25–93) for CLD to 914 (95% CI, 564–1122) for IHD. The proportion of RSV-RTI hospitalizations ranged from 3.6% (95% CI, 1.4–5.0) for adults with CLD to 7.4% (95% CI, 4.5–8.4) with stroke (Supplementary Table 4). Estimates from the regression models generally fitted the observed data well (Supplementary Figures 1 and 2). In comparison, the average annual estimates of influenza-associated RTI hospitalization ranged from 77 (95% CI, 32–80) among adults with CLD to 888 (95% CI, 371–1073) with IHD (Supplementary Table 3).

Average Annual Rate and RR of RSV-Associated RTI Hospitalization

RSV-RTI hospitalization rates in adults aged ≥45 years with comorbidity varied between Denmark and Scotland (Table 1). In Denmark, we found the highest risk in adults with CKD, with a rate of 9.4 (95% CI, 18.9–25.7) RSV-RTI hospitalizations per 1000 individuals per year and an RR of 7.2 (95% UR, 5.2–10.3) as compared with the Danish overall population aged ≥45 years. For the other comorbidities, we estimated a 4-fold higher rate in Danish adults with COPD and IHD vs the overall population, a 2-fold higher rate in those with diabetes and stroke, and a 1.5-fold higher rate in those with asthma. In Scotland, the highest risk was in adults with COPD, with a rate of 7.1 (95% CI, 4.6–8.8) RSV-RTI hospitalizations per 1000 individuals per year and an RR of 5.9 (95% UR, 4.1–8.7) vs the overall population aged ≥45 years. For the remaining comorbidities, we estimated a 4-fold higher rate in adults with IHD as compared with the overall population; a 3-fold higher rate in those with asthma, stroke, and CKD; and a 2-fold higher rate in those with CLD and diabetes. In comparison, rates of influenza-associated RTI hospitalization ranged from 4.4 (95% CI, 1.8–4.6) per 1000 adults with CLD per year to 7.6 (95% CI, 3.7–8.1) for those with asthma (Supplementary Table 3).

An exploratory analysis showed that in Denmark, the estimated yearly RSV-RTI hospitalization rates in adults aged ≥45 years with comorbidity revealed an increasing trend between the 2013–2014 and 2017–2018 seasons (Supplementary Figure 3). In the contrast, the rates remained similar in Scotland from the 2010–2011 to 2015–2016 seasons (Supplementary Figure 4).

RSV-RTI Hospitalization Rates in Adults With Comorbidity by Age Group

In the 2 nations, hospitalization rates rose quite steeply in the age groups spanning the 60s to 80s and were highest in those aged ≥85 years across most of the comorbidity groups (Table 2). In particular, adults with COPD and asthma had >6-fold higher hospitalization rates at age ≥75 years as compared with their younger counterparts in the 2 nations. The age-related patterns were less profound for the other comorbidities.

Across age groups, adults with comorbidity had higher rates than the overall population with an RR estimate >1.0, except for a few occasions (Figures 1 and 2, Supplementary Table 6). The RR estimates for adults with comorbidity vs the overall population were generally higher in those aged 45 to 54 years and/or 55 to 64 years. In particular, Danish and Scottish adults aged 45 to 54 years with IHD had RR estimates of 7.8 (95% UR, 6.0–10.3) and 6.0 (95% UR, 2.5–15.6), respectively; Scottish adults aged 55 to 64 years with COPD had an RR estimate of 6.7 (95% UR, 3.7–12.7). Exceptions to this pattern were observed in adults with asthma (both countries) and stroke (Scotland only). In the 2 countries, adults with asthma had a particularly high RR estimate at age ≥75 years, whereas Scottish adults with stroke had similar rates to the overall population at 45 to 64 years. Danish adults with diabetes had an RR of 4.8 (95% UR, 3.7–6.3) at 45 to 54 years but a RR <1.0 at ≥55 years; Scottish adults aged 65 to 74 years and younger had an RR estimate >1.0.

Sensitivity Analyses

Estimates of RSV-RTI hospitalization rates in the sensitivity analyses were broadly comparable to the main analyses in 2 countries (Supplementary Table 7). Scottish estimates of rates in adults aged ≥45 years with comorbidity, based on Poisson regression models and the models with the addition of rhinovirus-positive cases, changed by around ≤10% as compared with those in the main models. Danish estimates for adults with COPD, IHD, and asthma based on Poisson regression models were about 13% to 17% lower than the main analyses; the estimates for the other comorbidities changed only marginally (<10% changes). Models with zero lags in RSV and influenza predictors showed lower accuracy when compared with the main models, yielding comparable estimates for RSV in the Scottish population and lower estimates in the Danish population.

DISCUSSION

Using national hospital and virologic databases, this study provides estimates of RSV-associated RTI hospitalization among adults aged ≥45 years with 7 comorbidities and their RRs vs the overall population (with or without comorbidities) in Scotland and Denmark. We found that in Denmark, annual RSV-RTI hospitalization rates per 1000 adults aged ≥45 years ranged from 3.1 for asthma to 19.4 for CKD; in Scotland, the rate ranged from 2.4 for CLD to 9.0 for COPD. Adults with the comorbidities showed a 1.5- to 7-fold higher rate when compared with the overall populations aged ≥45 years across 2 countries. Adults with COPD, IHD, and diabetes consistently showed over 4-, 3-, and 2-fold higher rates in 2 countries, while more between-country differences were found for asthma and CKD. Across age groups, adults with comorbidity had higher rates vs the overall populations of same age except for a few occasions. Adults aged ≥65 years with the comorbidities (except for Danish adults with diabetes) have a substantially high burden of RSV hospitalization, especially among those with COPD or asthma.

Our estimates of RSV-RTI hospitalization rates in adults with COPD are comparable to a US study that reported a rate of 13 per 1000 individuals per season in adults aged ≥65 years with COPD or chronic heart failure, though greater than the results by a New Zealand study [30]. In the New Zealand study, the RSV-RTI hospitalization rate was between 0.2 and 1.4 per 1000 adults aged ≥50 years with COPD, asthma, congestive heart failure, coronary artery disease, cerebrovascular accident, diabetes, or end-stage renal disease. The disparities in estimates of RSV-RTI hospitalization rates could be related to the use of different analytic approaches. In contrast to our study using a regression modeling approach, the New Zealand study reported laboratory-confirmed RSV hospital cases.

The RR estimates reflect the risk of RSV-RTI hospitalization associated with the comorbidities and are broadly comparable to previous population-based studies. A modeling study in England showed that adults aged ≥65 years with a range of comorbidities were 4-fold more likely be hospitalized for RSV-associated respiratory diseases when compared with those without comorbidities [31]. In New Zealand, adults with 7 comorbidities had between 2- and 10-fold higher rates vs those without the comorbidities, and the RR varied across age groups and comorbidities [11]. Similar to our results, the New Zealand study found a smaller increase in the risk of RSV-RTI hospitalization in adults with diabetes as compared with chronic respiratory and heart diseases, and the risk elevation was mainly observed in young and middle-aged adults with diabetes and was less notable in older adults [11]. Our RR estimates should be viewed as a conservative estimation of the risk in adults with comorbidity in Denmark and Scotland, as we used the total population, with or without comorbidities, as a control.

The age-specific estimates of RSV hospitalization rates and RRs provide evidence on high-risk population groups that is relevant for RSV immunization recommendations and participant selection for vaccine evaluation in clinical trials and real-world studies. Besides, by using overall adults aged 65 to 74 years as a benchmark, estimates of RSV hospitalization rates among adults with certain comorbidities remained elevated at 45 to 54 years (COPD, asthma, and IHD) or 55 to 64 years (COPD, asthma, IHD, and CKD). The point estimates in adults with CKD aged 45 to 54 years were higher than the Scottish overall population aged ≥65 years, though it was difficult to draw a conclusion due to the wide confidence intervals of the estimates.

We found that RSV-associated RTI hospitalization rates in Scottish adults aged ≥45 years with the comorbidities were similar to those of influenza (Supplementary Table 5). Findings of the comparison could assist health authorities in RSV vaccination policy making in the presence of risk-based recommendations on influenza vaccination. Bruyndonckx et al reported that, unlike influenza, risks of unresolved symptoms and illness deterioration are associated with increasing age for RSV and that patients aged >75 years are at increased risks, further highlighting that prevalence and illness course significantly worsen at higher ages for RSV whereas only illness course is affected for influenza [32].

There are limitations and challenges in our study. One challenge in analyses of adults with comorbidity relates to the difficulty in defining and identifying comorbidities [11, 26]. In this study, we identified comorbidities that were recorded on any occasions in the national hospital care registries within 5 years before an RSV hospital episode, by assuming that preexisting comorbidities were recorded in any diagnostic fields at least once during recent health care utilizations prior to the RSV episode. It is possible that some individuals, especially adults at younger ages, were not diagnosed because they had been at the early stage of disease progression, had mild symptoms, or had remained good health status for ≥5 years. Given these factors and possible incomplete recording of comorbidities in the databases, our study provides a conservative estimate of true RSV-associated RTI hospital burden in adults with the comorbidities. Second, we assume that the virologic data (RSV and influenza) in the overall population are an indicator of viral activity in the adults with comorbidity. The assumption is based on observations that the time series data of hospitalized RTI cases and the virologic data had similar trends and times of peaks (Supplementary Figures 1 and 2). Where they existed, variations between the comorbidities and age groups were accounted for via the independent modeling process—namely, in RTI hospitalization rates and potential temporal sequential patterns in viral circulation [26]. Moreover, the model fits and results from sensitivity analyses suggest that our estimates of RSV-RTI hospitalization rates are generally robust.

The RSV testing level appears stable during the study period, yet variation in other patterns of testing (eg, age structure) over the years might affect our estimates. Third, we were unable to provide precise age-specific estimates for Scottish adults with CLD or Danish adults with CKD, due to the small RTI case counts in the groups. Finally, some individuals might have >1 comorbidity, which could cause biases in the estimates. Future research could compare the role of multiple comorbidities on risk of RSV infection. It is important to recognize the profound effect of COVID-19 and the impact of vaccination on RSV epidemiology and seasonality, which will likely confound any modern attempts to define RSV burden, especially in high-risk populations.

CONCLUSION

Using a standardized approach on national hospital and virologic databases, we provide age-specific estimates of RSV-RTI hospitalization among adults aged ≥45 years with 7 comorbidities in 2 European countries. Our results show that adults aged ≥65 years with these comorbidities and those aged 45 to 64 years with COPD, IHD, asthma, or CKD remain a high-priority population to consider for RSV immunization when RSV vaccines become available for use in the adult population [12, 13].

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Acknowledgments. We acknowledge the support of the eDRIS team (electronic Data Research and Innovation Services) at Public Health Scotland for its involvement in obtaining approvals, provisioning and linking of datasets, and the use of the secure analytic platform with the National Safe Haven. We acknowledge the significant contributions of the late Dr John Paget to this study.

PROMISE investigators. Harry Campbell and Harish Nair (University of Edinburgh), Hanna Nohynek (THL, Finland), Anne Teirlinck and Michiel van Boven (RIVM, the Netherlands), Terho Heikkinen (University of Turku and Turku University Hospital, Finland), Louis Bont (University Medical Center Utrecht), Peter Openshaw (Imperial College London), Phillipe Beutels (University of Antwerp), Andrew Pollard (University of Oxford), Veena Kumar (Novavax), Alexandro Orrico Sánchez (FISABIO, Spain), David Gideonse (RIVM, the Netherlands), Tin Tin Htar (Pfizer), Charlotte Vernhes (Sanofi Pasteur), Gael Dos Santos and Rachel Cohen (GlaxoSmithKline), Jeroen Aerssens (Janssen), Rolf Kramer and Ombeline Jollivet (Sanofi Pasteur), Nuria Manchin (TEAMIT).

Author contributions. H. C. and H. N. conceptualized the study. X. W. and R. O.-Y. analyzed the Scottish data. C. K. J. and A. M. E.-C. analyzed the Danish data. J. C. prepared figures and tables. R. O. Y. and X. W. led the interpretation and wrote the draft with important inputs from H. N., H. C., T. L., A. U. F., J. P., T. K. F., C. K. J., A. M. E.-C., and J. C. All authors critically reviewed the manuscript.

Disclaimer. This publication reflects only the authors’ view, and the Innovative Medicines Initiative 2 Joint Undertaking is not responsible for any use that may be made of the information that it contains herein.

Financial support. This work was supported by funding from the Innovative Medicines Initiative 2 Joint Undertaking (grant 101034339 to PROMISE); Nanjing Medical University Talents Start-up Grants (grant NMUR20210009 to X. W.); and the European Union's Horizon 2020 Research and Innovation Programme and the European Federation of Pharmaceutical Industries and Associations (to the Innovative Medicines Initiative 2 Joint Undertaking).

Supplement sponsorship. This article appears as part of the supplement “Preparing Europe for Introduction of Immunization Against RSV: Bridging the Evidence and Policy Gap.”

References

Author notes