The aim of the current study was to estimate influenza- and respiratory syncytial virus (RSV)-associated mortality and hospitalisations, especially the influenza-associated burden among low-risk individuals ≤65 yrs old, not yet recommended for influenza vaccination in many European countries.
Retrospectively during 1997–2003, Dutch national all-cause mortality and hospital discharge figures and virus surveillance data were used to estimate annual average influenza- and RSV-associated excess mortality and hospitalisation using rate difference methods.
Influenza virus active periods were significantly associated with excess mortality among 50–64-yr-olds and the elderly, but not in younger age categories. Influenza-associated hospitalisation was highest and about equal for 0–1-yr-olds and the elderly, and also significant for low-risk adults. Hospitalisation among children was mostly due to respiratory conditions, and among adults cardiovascular complications were frequent. RSV-active periods were associated with excess mortality and hospitalisation among the elderly. The highest RSV-related excess hospitalisation was found in 0–1-yr-olds.
Influenza-associated mortality was demonstrated in 50–64-yr-olds. Among low-risk individuals ≤65 yrs of age, influenza-associated hospitalisation rates were highest for 0–4-yr-olds, but also significant for 5–64-yr-olds. These data may further support extension of recommendations for influenza vaccination to include younger low-risk persons. The respiratory syncytial virus-associated burden was highest for young children but also substantial for the elderly.
Almost yearly, the influenza virus is held accountable for large numbers of deaths and hospitalisations 1–3, in particular among the elderly and people with high-risk medical conditions. Therefore, most countries recommend influenza vaccination for these groups 4. Recently, the USA extended vaccination to low-risk 50–64-yr-olds and young children, and in Canada vaccination for all ages was introduced 5, 6. Many European countries are now considering extending recommendations for influenza vaccination. More information, however, is needed about the potential impact of such changes in vaccination policy. In particular, figures of influenza-associated hospitalisation among low-risk adults are scarce 7.
It is difficult to estimate the influenza-associated healthcare burden accurately, because influenza virus infections are generally not virologically confirmed and are often not recognised clinically 8, 9. In addition, the influenza virus infection may predispose to other conditions, such as bacterial superinfection and cardiovascular complications 10–12. Co-circulation of other respiratory viruses during influenza season, in particular the respiratory syncytial virus (RSV) 13, makes it challenging to estimate the influenza-associated burden indirectly. Several studies suggested that RSV is responsible for considerable morbidity and even mortality not only in children but also among older adults 2, 14–16. Over the last 10 yrs the development of vaccines against RSV has progressed 17, and although a vaccine is not expected in the very near future, insight into the RSV-associated healthcare burden would be valuable.
Contrary to former studies 3, 7, viral surveillance data in the Netherlands from 1997–2003 revealed largely separate peaks of influenza virus- and RSV-activity that allowed quantification of the impact of both viruses separately. The aim of the current study was to assess influenza- and RSV-associated mortality and hospitalisation, especially the influenza-associated burden among low-risk individuals ≤65 yrs of age.
During the period 1997–2003, laboratory-based surveillance for various viruses was conducted by the Weekly Sentinel System of the Dutch Working Group on Clinical Virology in the Netherlands. A group of 17 virological laboratories reported weekly on the absolute number of patients, either hospitalised or visiting outpatient clinics, who tested positive for a certain virus. Surveillance data for influenza virus and RSV from that system were used in the current study. Most of the laboratory diagnoses of influenza virus and RSV infections were made by virus isolation on cell culture or rapid antigen tests. The weekly virological reports were demonstrated to adequately reflect trends in national viral activity 18. However, most RSV surveillance data (96%) were reported in children <5 yrs old 19. An influenza virus subtype was considered dominant when it accounted for ≥50% of all isolates that were subtyped in that season. The influenza virus and RSV surveillance data are summarised in table 1⇓. All influenza seasons were subtype A H3N2-dominant, except for 2000–2001, which was subtype A H1N1-dominant.
Definition of study periods
With minor modifications, study periods were defined according to Izurieta et al. 20. For each winter season, from week 40 of 1 yr to week 20 of the next, the influenza virus-active period was defined as the periods of at least two consecutive weeks in which each week accounted for ≥5% of the season's total number of laboratory-confirmed influenza cases 20. Similarly, the RSV-active period was defined as the period of at least 2 consecutive weeks in which each week accounted for ≥5% of the season's total number of RSV-positive patients. The period with influenza predominance was defined as the influenza virus-active weeks with <5% of the season's total number of positive tests for RSV 20. The peri-seasonal baseline period was defined as periods of at least two consecutive weeks within week 40–20 in which each week accounted for <5% of the season's total number of influenza and RSV-positive cases. The summer baseline period was defined as week 21–39. Unlike the study of Izurieta et al. 20, the weeks in which para-influenza virus was isolated were not excluded from the study as (sporadic) isolates were reported throughout the year. For the same reason, weeks in which sporadic isolates of the influenza virus and RSV were reported during summer baseline period were not excluded from the study.
During the study period there were 92 influenza and/or RSV-active weeks; 46 weeks of influenza predominance, 42 weeks of RSV predominance, and only 4 weeks of both influenza virus- and RSV-activity.
Mortality data and outcomes
National weekly mortality figures were obtained from Statistics Netherlands (Voorburg/Heerlen, the Netherlands). No information about the presence of high-risk conditions was available in these figures. Weekly hospitalisation rates were provided by Prismant (Healthcare and Advice Institute, Utrecht, the Netherlands), who register all hospitalisations nationwide according to the International Classification of Diseases-9CM. In this register, all discharge diagnoses were registered per hospitalisation with the first diagnosis marked as primary diagnosis. During the study period, all hospitalisations with discharge diagnoses indicating acute upper respiratory disease (460–465, 381–384, 034), acute or chronic lower respiratory disease (466, 480–487, 490–496, 510–518, 78609, 7862), cardiovascular disease (410–415, 420–422, 428–429, 7852), cerebrovascular disease (431–437), bacterial invasive disease (036, 038, 041, 320, 3220, 3229, 7280, 7907) or other conditions possibly related to a respiratory infection were collected (293, 323, 390–392, 3483, 7803, 7806, 7784). Hospitalisations were divided into upper respiratory tract infections (URTI), lower respiratory tract infections (LRTI) and pulmonary disease, cardiovascular complications (CVC) and others (e.g. bacterial invasive disease, fever without focus and delirium). Apart from the discharge diagnosis, the date of hospitalisation, the age and the presence of high-risk conditions were registered. A high-risk condition was considered present when at least one of the 14 subdiagnoses indicated chronic respiratory disease (491–496, 500–508, 516–518, 5199, 71481), chronic cardiac disease (391, 393–396, 402, 404, 410–412, 414, 416, 424–429, 745–747), diabetes mellitus (250–251), renal insufficiency (581–591), haematological malignancy (2031, 2038, 204–208) or HIV/AIDS (042–044). When a chronic cardiac or respiratory condition was marked as primary discharge diagnosis, this was also considered as the presence of a high-risk condition.
The population of each consecutive year on January 1st was taken as the population at risk, assuming a stable population throughout the year (Statistics Netherlands). For all years taken together, the average weekly mortality rate and rate of hospitalisation (per 100,000 subjects) was calculated in different study periods, i.e. peri-seasonal and summer baseline periods and periods in which influenza virus or RSV predominated. Weekly excess mortality and hospitalisations with 95% confidence intervals (CIs) associated with influenza virus and RSV were determined by subtracting summer and peri-seasonal baseline rates from rates during periods of influenza virus or RSV predominance. The cumulative annual winter excess rate was the total excess per 100,000 subjects associated with influenza virus or RSV during winter season, and was calculated by multiplying the average weekly excess rate during the influenza predominance period with the number of influenza virus-active weeks during that winter season. The excess rates were applied to the national population of 2005 (Statistics Netherlands) to estimate the total number of deaths and hospitalisations associated with influenza virus and RSV in the Netherlands. The proportions of the population with high-risk disease, i.e. medical conditions which are associated with a higher risk of complicated influenza virus infections, were obtained from the National Information Network Primary Care 21. Since the prevalence of high-risk disease among children was relatively low, no subset analysis was performed according to the presence of high-risk disease among children. Subset analysis according to the presence of high-risk disease was also not performed for subjects aged ≥ = 65 yrs, as these subjects are already recommended for influenza vaccination.
In total, 839,303 all-cause deaths and 1,551,598 hospitalisations for URTI, LRTI, CVC and others were registered. Of these all-cause deaths, 1% was reported in 0–17-yr-olds, 6% in 18–49-yr-olds, 13% in 50–64-yr-olds and 80% in those aged ≥65 yrs. For hospitalisations, these figures were 14, 12, 23 and 51%, respectively.
No evident excess mortality was found in the age categories 0–1, 2–17 and 18–49 yrs during influenza virus-active periods (table 2⇓). However, among those aged ≥50 yrs, significant influenza-associated excess mortality was recorded. Among 50–64-yr-olds, influenza-associated excess mortality was highest for 60–64-yr-olds (fig. 1⇓). Influenza-associated excess hospitalisation was highest in 0–1-yr-olds (table 3⇓). Infants appeared responsible for the largest part of this excess hospitalisation for LRTI, namely a yearly average of ∼13–221 hospitalisations per 100,000 infants <1-yr-old (respectively with the peri-seasonal and summer baseline period as reference) and 13–64 hospitalisations per 100,000 1-yr-olds. In adults, significant excess hospitalisation for LRTI and CVC was recorded during influenza virus-active weeks (table 4⇓). Excesses for all diagnosis categories increased with age and increased among low-risk 50–64-yr olds (fig. 2⇓). In absolute numbers, the highest influenza-associated healthcare burden occurred in the elderly (fig. 3⇓).
During RSV-active periods, no evident excess mortality was found in the age categories 0–1, 2–17 and 18–49 yrs (table 2⇓). The youngest children appeared to experience the largest RSV-associated excess hospitalisation for LRTI (table 3⇓), with average annual hospitalisation ∼870–1063 per 100,000 infants <1-yr-old (with respectively peri-seasonal and summer baseline period as reference) and 104–151 per 100,000 1-yr-olds. Significant excess hospitalisation was recorded in adults during RSV-active periods, in particular in the elderly (table 4⇓). The total absolute number of RSV-associated excess hospitalisation was highest and approximately similar among 0–1-yr-olds and the elderly (fig. 4⇓).
This nationwide retrospective study covering six recent consecutive respiratory seasons showed that mortality associated with influenza was substantial among those aged ≥50 yrs. Influenza-associated hospitalisation was significant among healthy persons of all age categories and highest for young children and older people. The highest RSV-associated excess hospitalisation also occurred in the youngest age group, but was also significant in the elderly in which RSV-active periods were also associated with excess mortality.
Many models have been described to estimate the influenza-associated burden, and most are based on determining the excess rate during influenza virus-active periods versus baseline periods with lower or no influenza virus-activity. The rate–difference model has regularly been applied 7, 20, 22 and a straightforward variant of these models allowing for insight to a broad public. Due to the use of diverse statistical models, including the different definitions of viral seasons and the various definitions of end-points (e.g. culture-confirmed or nonconfirmed influenza), studies are difficult to compare 1–3, 7, 9, 20, 22–28. Variations of the included study period (and consequently varying influenza virus-activity) and differences in healthcare systems further lead to poor comparability, for example primary care in the Netherlands with a gate-keeping function may affect hospitalisation rates.
In contrast to previous studies 2, 26, which reported annual influenza-associated deaths of 2–7 per 100,000 among 0–1-yr-olds and ∼1 per 100,000 among 1–4-yr-olds, the current authors could not detect excess mortality in children during influenza virus-active periods. The present methods may lack sensitivity to detect small excesses of influenza-associated deaths. However, the current study confirmed that among children and 18–64-yr-olds without high-risk medical conditions, the highest influenza-associated excess hospitalisation occurs in the youngest children 3, 9, 20, 22–25. This suggests that this target group may benefit particularly from influenza vaccination, certainly when influenza-related primary care visits and parental work absenteeism are also taken into account 28. Furthermore, it is also thought that children are the main disseminators of influenza 29, and vaccinating children may therefore limit the spread of infection in the community. The influenza vaccine is, however, currently not licensed for children aged <6 months, and evidence for the efficacy and effectiveness of the vaccine in children under 2 yrs of age is limited 30.
The present study indicates that influenza virus-active periods were associated with excess mortality among 50–64-yr-olds. Unfortunately, the study was not able to estimate which part of this excess occurred in low-risk individuals, as information about risk status was not available in the mortality figures. Influenza-associated hospitalisation was, however, significant among low-risk 50–64-yr-olds. A recent study was not able to demonstrate influenza-associated hospitalisation in this group, but this was probably due to limited statistical power 7. The hospitalisation rates found in the current study among low-risk 50–64-yr-olds were clearly lower than those in young children, but the nature of the hospitalisations may also be important. While in children the excess hospitalisation was mainly due to respiratory conditions, hospitalisations for CVC made up the largest part of the excess hospitalisation among 50-64-yr-olds. Obviously, these hospitalisations are expected to have a large impact on the healthcare system and financial resources. Both influenza-associated excess mortality and hospitalisation, in particular for CVC, increased with age, indicating that 60–64-yr-olds would benefit most from annual influenza vaccination. The influenza vaccine has proven to be safe and effective among adults, and it also appears effective in preventing cardiovascular outcomes 31–34. Apart from the potential health gain, cost-effectiveness analyses taking into account both direct and indirect influenza-associated costs, like absenteeism from work, are important to direct decisions to change vaccination policy.
Despite the high influenza vaccination coverage among the elderly in the Netherlands (70–80%), influenza-associated mortality appeared high among the elderly in the present study. It is known, however, that the immunogenicity of the influenza vaccine decreases with age after the age of 65 yrs, which may lead to reduced effectiveness 31, 35. This emphasises the need for further improvement of the protection against influenza, particularly in the elderly.
As expected, the highest RSV-related excess hospitalisation occurred in the youngest age group 36, 37, and this burden appeared considerably higher than that associated with influenza. The RSV-related mortality in this age category could not be demonstrated. Due to the same power problems applicable to the influenza-related mortality among young children in the current study, the possibility of RSV-associated mortality in this age category could not be excluded. Previous studies reported annual RSV-associated deaths of 5–8 per 100,000 among infants aged 0–12-months and ∼1 per 100,000 among 1–4-yr-olds 2, 26. Significant excess hospitalisation was also demonstrated among adults, especially the elderly. Moreover, RSV-active periods appeared associated with excess mortality among 50–64-yr-olds and the elderly. The main RSV-associated hospitalisations were for respiratory indications and, to a lesser extent, for CVC compared with influenza-associated hospitalisation, which is in agreement with a former study 25.
To appreciate the results of the current study, some aspects should, however, be discussed. As epidemiological data were used to estimate influenza- and RSV-related burden, direct evidence was lacking for the causative pathogen that led to hospitalisation or death. Therefore, the results should be interpreted cautiously. The burden will, however, be underestimated by only recording laboratory-confirmed influenza and RSV-infections, due to underdiagnosis and under-reporting, but also in the case of secondary complications (like bacterial infections or other possible complications such as cardiovascular diseases). Moreover, the influenza virus is a pathogen that has been extensively studied and is known to be responsible for considerable annual morbidity and mortality. In contrast, the role of RSV in causing morbidity and mortality, especially among adults, is less clear. In the current study, a clear excess mortality and morbidity was found during RSV-active periods. However, most of the present RSV surveillance data were reported in children, and although the present authors assumed that RSV-activity among adults parallels that in children, this has not yet been suitably proven 38. It is possible therefore, that some of the RSV- and influenza-associated morbidity could have been misclassified, as in all other excess studies. This stresses the importance for age-specific RSV-surveillance and should be addressed in future studies.
The estimations of virus-related burden strongly depended on the applied reference period, and estimates should, therefore, be viewed in rather large margins. The peri-seasonal baseline period is the most conservative reference and the application of this probably underestimated the virus-related burden, because excess rates are determined over periods in which influenza virus and RSV are active albeit to a lesser extent (weeks with <5% of season's total number of isolates). Conversely, the potential role of other respiratory viruses or other seasonal factors, such as certain meteorological conditions affecting the rate of hospitalisation and mortality, is limited with the peri-seasonal baseline period as reference. In other words, by applying the peri-seasonal baseline period as reference, the present authors attempted to correct for other potentially important seasonal factors. Therefore, it appears that the true influenza- and RSV-associated excess mortality and hospitalisation probably lies within the estimations based on the peri-seasonal and summer baseline period. Nevertheless, it is expected that other viruses like rhinoviruses and coronaviruses cause milder clinical manifestations of respiratory infections, which lead to primary care visits. Moreover, surveillance data in the Netherlands during the current study period demonstrated that rhinoviruses, adenoviruses and para-influenza viruses appeared to have no clear seasonal pattern like influenza viruses and RSV, with rather long periods of marginally increased activity or very short peaks of increased activity (see Appendix). Unfortunately, the seasonal pattern of some recently discovered coronaviruses and the human metapneumovirus could not be assessed as no surveillance data were available during the study period.
The major strength of the current study is the nationwide inclusion of large numbers, thus allowing subanalysis according to age and the presence of high-risk conditions for hospitalisation among adults. With the Netherlands being a small but densely populated country, population characteristics are relatively homogenous nationwide and viral circulation is more or less simultaneous across the country, making ecological studies more reliable. Furthermore, the study period covered 6 yrs with different viral attack rates.
In summary, substantial influenza-associated excess hospitalisation was found among 0–4-yr-olds, although mortality could not be attributed to influenza in this age group. Among low-risk 50–64-yr-olds, significant influenza-associated excess hospitalisation was also recorded, and even excess mortality appeared to be present. Part of this burden might be prevented by the introduction of an annual influenza vaccination. The respiratory syncytial virus-associated burden appeared substantial, particularly in young children but also in the elderly, and therefore the role of a future respiratory syncytial virus vaccine appears promising in reducing this healthcare burden.
Statement of interest
This study was funded by Dutch Health Council and The Netherlands Organization for Health Research and Development ZONMW (grant number 2200.0121).
For editorial comments see page 1029.
- Received March 22, 2007.
- Accepted August 1, 2007.
- © ERS Journals Ltd