Background: 600 000 deaths worldwide are estimated to be directly or indirectly attributable to respiratory syncytial virus (RSV).
Objectives: To determine: (1) the mortality rate; and (2) risk factors for death in children with severe RSV infection.
Setting: 20-bed, regional, multidisciplinary, tertiary, paediatric intensive care unit (PICU) in a university-affiliated children’s hospital.
Methods: Cohort study of all children with severe RSV infection covering eight consecutive RSV seasons (1999–2007), using PICU admission as a marker of severity.
Results: Of the 406 RSV-positive patients that were admitted to PICU: 98.5% required mechanical ventilation; 35 children died—median age 5.1 months (interquartile range (IQR) 2.4–13.6), length of PICU stay 16 days (IQR 8–31) and 371 survived—median age 2.5 months (IQR 1.2–9), length of PICU stay 5 days (IQR 4–9). The overall PICU RSV mortality was 8.6% with a standardised mortality ratio of 0.76. During the study period 2009 RSV-positive patients were admitted to the children’s hospital, giving a hospital RSV mortality rate of 1.7%.
Of the deaths, 18 were directly RSV related (RSV bronchiolitis-related mortality PICU 4.4% and hospital 0.9%) as the patients were still RSV positive when they died and 17 children died from non-pneumonitis causes after becoming RSV negative.
All of the RSV deaths had pre-existing medical conditions — chromosomal abnormalities 29%, cardiac lesions 27%, neuromuscular 15%, chronic lung disease 12%, large airway abnormality 9%, and immunodeficiency 9%. Nineteen children (56%) had pre-existing disease in two or more organ systems (relative risk (RR) 4.38).
Predisposing risk factors for death were pre-existing disease (RR 2.36), cardiac anomaly (RR 2.98) and nosocomial/hospital-acquired RSV infection (RR 2.89). There is an interaction effect between pre-existing disease, nosocomial/hospital-acquired RSV infection and mortality (p<0.001).
Conclusions: Pre-existing disease/comorbidity, in particular multiple pre-existing diseases and cardiac anomaly, is associated with a significantly higher risk of death from severe RSV infection. Nosocomial/hospital-acquired RSV infection is an additional major risk factor for death in children with severe RSV infection.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Worldwide each year 600 000 deaths occur that are directly or indirectly attributable to respiratory syncytial virus (RSV).1 It has been estimated that in children under 1 year of age nearly 50 deaths per year in the UK2 and over 200 deaths in the USA3 are attributable to RSV. RSV has been found to be the most common viral cause of death in children below 5 years of age and especially those less than 1 year old.3 The estimated paediatric all-cause RSV mortality rates were 8.4 (UK) and 5.4 (USA) per 100 000 population in the less than 1 year old age group, and decreased substantially with increasing age.2 3 However, when restricted to respiratory causes, only the UK rate fell to 2.9 deaths per 100 000 population in the less than one year old age group.2 Epidemiological studies have demonstrated that the bronchiolitis-related mortality rate has remained low and stable in the USA, and has declined in the UK over the past two decades.4 5
It is recognised that a considerable proportion of children who die from bronchiolitis have serious underlying conditions/comorbidity, such as cardiac and central nervous system anomalies, chromosomal abnormalities and immunodeficiencies.4 5 However, these pre-existing conditions are often not effectively identified as contributing factors in epidemiological studies that rely on national statistics coding.2–6 8 9
In order to determine the mortality rate and risk factors for death in children with severe RSV infection, a study of all children admitted for paediatric intensive care over an 8-year period (ie, eight RSV seasons) was performed.
Twenty-bed, regional, multidisciplinary, tertiary, paediatric intensive care unit (PICU) in university-affiliated children’s hospital. The PICU has an annual admission rate of over 1000 children. Cardiac and medical patients each account for 40% of the PICU admission and 20% are surgical (all subspecialities represented).
To determine: (1) the mortality rate; and (2) risk factors for death in children with severe RSV infection.
All children with RSV bronchiolitis, confirmed on RSV antigen testing, immunofluorescence and/or culture that required PICU admission between June 1999 and May 2007 (ie, encompassing eight consecutive RSV seasons — October to March) were studied.
The data were collected prospectively during the last six RSV/winter seasons (October 2001 onwards) and retrospectively for the first two seasons. PICU admission was utilised as the marker of severity of disease.
It is national policy that all children who require intensive care and mechanical respiratory support are transferred to the regional PICU. All forms of mechanical ventilation (conventional and high frequency) and continuous positive airways pressure (CPAP) support are carried out on the PICU. Muscle relaxants are not routinely used. Non-invasive CPAP support is usually performed on our 14-bed high dependency unit (HDU), which is separate from the PICU. More recently some of the regional secondary-level hospitals have been performing non-invasive CPAP support within their paediatric HDU/wards.
Children requiring extracorporeal life support (ECLS)/extracorporeal membrane oxygenation (ECMO) are transferred to one of four national centres.
Diagnostic samples of nasopharyngeal aspirates (for RSV detection) and lower airway secretions (for bacterial culture and viral detection) were taken on admission to PICU and thereafter when clinically indicated. From January 2002 onwards RSV surveillance samples were taken on all children during the RSV season (October–March) on admission and twice a week while on PICU. Samples were collected by specialist respiratory physiotherapists or PICU staff members.
Viral–nasopharyngeal aspirates were tested by the Directigen RSV test (Becton Dickinson microbiology systems, Maryland, USA) (1999–2004) or NOW RSV test (Binax Inc., Maine, USA) (2005–2007). These are in vitro immunochromatographic assays for the rapid and qualitative detection of RSV antigen directly from nasopharyngeal specimens. Samples negative for RSV using the rapid antigen tests were cultured using standard virological techniques at the Public Health Laboratory (Aintree, Liverpool, UK). Immunofluorescence for RSV was performed using standard virological techniques at the Public Health Laboratory.
Bacterial/yeast-diagnostic or clinical samples were processed immediately in a qualitative and semi-quantitative way using standard microbiological methods. Standard methods for identification, typing, and sensitivity patterns were used for all microorganisms.15
The empirical use of antibiotics in children with severe bronchiolitis was at the discretion of the attending consultant. Patients with signs of infection received intravenous cefotaxime (50 mg/kg/dose four times daily) as first-line therapy for 48 h while awaiting culture results. Clinical status on presentation governed whether supplementary intravenous cover with an aminoglycoside, gentamicin (7 mg/kg/day single daily dose) was added. Antibiotics were rationalised or stopped once culture and sensitivity results became available.
Ribavirin therapy is only considered in immunocompromised children.
Nosocomial RSV infection — when a child admitted to the PICU was RSV negative or from whom no samples were taken as they did not exhibit signs of bronchiolitis, and who then became RSV positive 5 or more days after the admission.16
Results were expressed as a percentage of the total study population; median and interquartile ranges (IQRs) were used to describe the demographic distributions. Prediction of mortality using paediatric index of mortality (PIM) was obtained on the patient’s first contact with the PICU team.17 Standardised mortality ratio (SMR) is the ratio of actual deaths compared to the number predicted (in this study derived from PIM score17). Continuous data were analysed using the Wilcoxon–Mann–Whitney (W–M–W) test. Categorical data were analysed using Fisher’s exact or McNemar’s test. Multivariate analysis was performed using linear and logistic regression analysis.
Statistical calculations were performed with the Statistical Program for Social Science release 11.0.0 (SPSS 11, Chicago, Illinois). A p value <0.05 was considered statistically significant. RR with 95% CI calculations was performed with confidence interval analysis software (CIA 2.1.2) from Altman DG, Machin D, Bryant TN, et al, eds. Statistics with confidence. London: BMJ Books, 2000.
During the study period 406 RSV-positive patients were admitted to PICU. The patient numbers and deaths for each RSV season are shown in table 1. The causes of death of the 35 RSV-positive children that died are shown in table 2. Overall PICU RSV mortality was 8.6% with an SMR of 0.76. During the study period 2009 RSV-positive patients were admitted to the children’s hospital, giving a hospital RSV mortality rate of 1.7%. Of the deaths 18 were directly RSV related as the patients were still RSV positive when they died. The remaining 17 children died from non-pneumonitis causes after becoming RSV negative on antigen testing and viral culture, giving a PICU RSV bronchiolitis-related mortality of 4.4% and a hospital RSV bronchiolitis-related mortality of 0.9%. RSV infection did not appear on the death certificate in 13 of the 35 deaths (37%).
Patient demographics of those admitted to PICU are shown in table 3. Mechanical ventilation was required in 96.5% of the children and non-invasive respiratory support only in 3.5%. Clinical characteristics of the children admitted to PICU are shown in table 4. Analysis of oxygen and ventilation indices,18 even though measured, was not undertaken as the high incidences of pre-existing cardiac anomalies and chronic lung disease in both groups would confound interpretation. Pre-existing diseases (chromosomal abnormalities, cardiac lesions, neuromuscular disorders, chronic lung disease, large airway abnormality, immunodeficiency) were present in 187 children (46%) on admission to PICU. Of these children with pre-existing disease, 142 (76%) were admitted primarily for RSV disease and the remainder were postoperative (cardiac and general surgery), neurology, chronic respiratory, or sepsis.
All of the RSV deaths had pre-existing medical conditions/comorbidity: chromosomal abnormalities 29%, cardiac lesions 27%, neuromuscular 15%, chronic lung disease 12%, large airway abnormality 9% and immunodeficiency 9%. Nineteen children (56%) had pre-existing disease in two or more organ systems and carried a significantly higher risk of death (table 4).
There was an interaction effect between pre-existing disease, nosocomial/hospital-acquired RSV infection and mortality (McNemar’s test p<0.001). Children with pre-existing disease and nosocomial/acquired RSV infection stayed in PICU longer (W–M–W test p<0.001) and were older (p<0.05). Pre-existing disease and nosocomial/hospital-acquired RSV infection were predictive of death by uni- and multivariate analysis (all p values <0.05), but age, gender, previous prematurity, bacterial co-infection and inter-hospital transfer did not add to their prognostic value. Children that died while RSV positive and those that died later once RSV negative were similar in age, age group, pre-existing disease (including cardiac and multiple organ system involvement), gender, bacterial co-infection, previous prematurity and inter-hospital transfer (Fisher’s exact test all p values >0.3). However, the latter group had a longer length of PICU stay — median (interquartile range (IQR)): 31 days (22–58) vs 11 days (8–17) (W–M–W test p<0.05).
Some of the data presented have been used in previous publications on the incidence of pulmonary bacterial co-infection in severe RSV bronchiolitis (three RSV seasons: 2002–2005)19 and a description of an RSV outbreak (RSV season 2002).14
Data on the prior receipt of prophylactic monoclonal RSV antibody therapy have not been reported as it was only introduced into clinical practice late in the study period and therefore any impact would be difficult to ascertain at this stage.
Three children received ribavirin (all were immunocompromised following chemotherapy). Surfactant therapy was utilised in two children with severe pneumonitis.
Five children received ECLS/ECMO, of which the four that died all had pre-existing disease, but not the survivor. One child that died on ECLS/ECMO had RSV and pertussis dual infection.
This observational cohort study over eight consecutive RSV seasons found that only children that had pre-existing disease/comorbidity died from severe RSV infection, that is, prior good health protected against death. PICU admission was utilised as a marker of severe RSV disease in this study, which differs from many other studies that defined hospitalisation as reflecting severe RSV disease.1 5 10 12 13 20 Having focused on a higher risk group than many of the previous reports, this study uniquely compared the risk factors in those that died with the survivors within this subgroup with severe RSV disease.
Even though these hospital-based data do not necessarily reflect incidences in the general community which are better reflected in epidemiological/population-based studies, they do reflect risk factors pertinent to RSV infection in the general paediatric population. The assessments and estimates from epidemiological studies are derived from data obtained from diagnosis-based coding categories retrieved from National Surveillance organisations (eg, Center for Disease Control and Prevention in USA and Health Protection Agency in UK). They therefore suffer from the acknowledged limitation of reliance on specific codes (eg, for respiratory, circulation deaths) which may not cover all RSV-associated pathologies.2–6 8 9 This would have been the case in a number of the deaths in our study population as their deaths occurred after becoming RSV negative and were not necessarily directly RSV related (table 1). The detrimental role or destabilising effect that the RSV infection played in the clinical course of these cases is difficult to assess accurately as the cause of death was often extra-pulmonary. Certainly death certificates did not report the recent RSV infection in nearly 40% of the deaths and this failure would contribute to the RSV influence being missed in epidemiological studies. This may have been a factor in the conclusions of Shay et al that the majority of RSV-related deaths did not occur in children at high risk for severe RSV disease, which is contrary to our data.6 Their USA study differed in that the study period was earlier (1979–1997) and its epidemiological design relied on cause-specific International Classification of Deaths 9th revision (ICD-9) codes on death certificates. Nearly all the children in the present study which died had well-recognised risk factors for severe RSV disease.10–13 20–24
In our institution RSV deaths in PICU reflect the total hospital deaths from RSV as it is highly unlikely that any children with severe RSV disease would not be admitted to PICU and more than 85% of hospital deaths occur in PICU. The hospital RSV mortality rate is similar to other studies,5 7 10 12 13 20 23 even taking into account that our children’s hospital is a regional referral centre, thereby collecting cases with severe illness. Additionally, our centre is the regional paediatric cardiac referral centre, resulting in children with congenital heart disease and bronchiolitis being more likely to be referred to our PICU for intensive care management. It is therefore not surprising that one-third of deaths in this RSV patient group were in children that were retrieved/transferred from other base hospitals. As found in other studies the number of deaths from RSV each year varied widely.2–5 Typing of the RSV strain is not performed in our institution, so we are unable to speculate as to whether the strain influenced mortality overall or season on season.
The most significant risk factor for death was pre-existing disease, of which a cardiac anomaly held a particularly high risk. Interestingly, contrary to what one might have presumed, cardiac lesions with high pulmonary blood flows (eg, endocardial cushion defects, ventriculoseptal defects, truncus arteriosus, etc) did not stand out from the rest as predisposing to mortality. Congenital heart disease being a significant risk factor for severe disease and death is in keeping with previous reports.6 8 10 11 13 23 Earlier studies have identified previous prematurity as a risk factor for hospital admission and death.5 10 13 20 24–26 However in our study population, when compared to the survivors with severe RSV disease, it was not an additional risk factor for death. Perhaps in this patient group previous prematurity predisposes to severe disease (morbidity), whereas other pre-existing conditions govern mortality. Having said this, caution should be exercised as the sample size may underpower the confidence of interpretations. The pre-existing disease most probably has bearing on the finding that, as a group, those that died were older and stayed in PICU longer.
Pre-existing disease may confound the relationship between nosocomial RSV infection and death. Children with more complex underlying conditions stay in hospital and on PICU longer and thereby have protracted exposure to potential cross-infection. This is borne out by the longer length of stay in the group of RSV deaths with its high incidence of both pre-existing disease and nosocomial RSV infection. Previous studies have described an increased incidence of nosocomial RSV infection in children with congenital heart disease, chronic lung disease and immunodeficiency.7 13 14 20 21 27 28 It highlights not only their vulnerability to nosocomial infection, but also its potential devastating impact on these children. Perhaps this subgroup of patients should be isolated (in a cubicle) and reverse barrier-nursed during the RSV season to minimise nosocomial infection, acknowledging that it may further stretch already strained isolation spaces. The prolonged days of RSV positivity in the group of RSV deaths may also relate to the high incidence of pre-existing diseases. It has been recognised that patients with underlying conditions may excrete virus/RSV for longer periods.14 21 22 27 28 So, in addition to susceptibility to nosocomial infection, once infected they constitute a prolonged source for potential cross-infection within the unit. RSV prophylaxis (expensive monoclonal antibody therapy) or an eagerly awaited vaccine (still under development) may offer some protection from RSV infection in the community for this vulnerable group of children with pre-existing disease.
Pre-existing disease/comorbidity, in particular multiple pre-existing diseases and cardiac anomaly, is associated with a significantly higher risk of death from severe RSV infection.
Nosocomial/hospital-acquired RSV infection is an additional major risk factor for death in children with severe RSV infection.
What is already known on this topic
In children with RSV bronchiolitis it is recognised that serious pre-existing disease/comorbidity—cardiac and central nervous system anomalies, chronic lung disease, chromosomal abnormalities, immunodeficiencies—is a risk factor for hospital admission and severe disease.
What this study adds
Pre-existing disease/comorbidity, especially cardiac anomaly and pre-existing diseases in two or more organ systems, hold a significantly higher risk of death from severe respiratory syncytial virus (RSV) infection.
Nosocomial/hospital-acquired RSV infection is an additional major risk factor for death in children with severe RSV infection.
Competing interests: None.
Ethics approval: The study was approved by the hospital’s Clinical Audit Department—a subgroup of the Research and Clinical Development Directorate.