|Home | About | Journals | Submit | Contact Us | Français|
Asthmatics hospitalised because of influenza A infection are less likely to require intensive care or die compared with nonasthmatics. The reasons for this are unknown.
We performed a retrospective analysis of data on 1520 patients admitted to 75 UK hospitals with confirmed influenza A/H1N1 2009 infection. A multivariable model was used to investigate reasons for the association between asthma and severe outcomes (intensive care unit support or death).
Asthmatics were less likely than nonasthmatics to have severe outcome (11.2% versus 19.8%, unadjusted OR 0.51, 95% CI 0.36–0.72) despite a greater proportion requiring oxygen on admission (36.4% versus 26%, unadjusted OR 1.63) and similar rates of pneumonia (17.1% versus 16.6%, unadjusted OR 1.04). The results of multivariable logistic regression suggest the association of asthma with outcome (adjusted OR 0.62, 95% CI 0.36–1.05; p=0.075) are explained by pre-admission inhaled corticosteroid use (adjusted OR 0.34, 95% CI 0.18–0.66) and earlier admission (≤4 days from symptom onset) (adjusted OR 0.60, 95% CI 0.38–0.94). In asthmatics, systemic corticosteroids were associated with a decreased likelihood of severe outcomes (adjusted OR 0.36, 95% CI 0.18–0.72).
Corticosteroid use and earlier hospital admission explained the association of asthma with less severe outcomes in hospitalised patients.
Viral respiratory tract infections are the commonest cause of exacerbations of asthma, being implicated in ~80% of exacerbations in children and 50% of exacerbations in adults [1–3]. The most commonly identified virus associated with exacerbations of asthma is rhinovirus, followed by coronavirus, influenza, parainfluenza and respiratory syncytial virus [1, 4].
During the 2009–2010 influenza pandemic, asthma was noted to be the commonest comorbid illness of patients admitted to hospital with influenza A/H1N1 (H1N1pdm) infection, present in ~25% of patients [5–7]. It was further observed that asthmatics admitted to hospital with influenza infection were less likely to die or require admission to intensive care compared with nonasthmatics [6, 8, 9]. In contrast, the presence of virtually all other chronic medical conditions was associated with an increased risk of death following hospitalisation . The reasons why a diagnosis of asthma should be associated with less severe outcomes in hospitalised patients have not been examined. Possible explanations suggested include: 1) a lower threshold for hospital referral and admission, hence a milder illness at time of admission compared with nonasthmatics; 2) the occurrence of treatable influenza-induced exacerbations of asthma rather than influenza-related pneumonia prompting admission; and 3) the tendency for asthma to be the dominant risk factor in younger age groups in whom the case-fatality rate due to influenza is lower [9, 10].
In May 2009, the UK Department of Health established the Influenza Clinical Information Network (FLU-CIN) to undertake clinical surveillance of hospitalised cases of H1N1pdm infection. The aim of this study was to provide rapid information on the admission of cases to hospital and to improve the management of influenza. We used the abundant information from this study to examine the risk factors and outcomes of persons with asthma.
Between April 2009 and January 2010, FLU-CIN collected clinical, epidemiological and outcome data on patients admitted to 75 hospitals in 31 UK cities or towns with confirmed H1N1pdm infection. The details of data collection and the overall findings from the first wave of the 2009 pandemic have been described elsewhere , but the analysis described here is based on the full dataset from the first and second waves combined . Briefly, H1N1pdm infection was diagnosed by a positive reverse transcriptase PCR result from respiratory samples obtained during the admission episode. A large dataset was collected by specifically trained data collectors, and included demography, pre-existing medical conditions, acute and long-term medications and clinical observations at the time of hospital admission.
A diagnosis of pneumonia was based on the criteria as previously described : 1) admission chest radiograph report clearly suggestive of pneumonia; 2) admission chest radiograph report of acute infiltrates but no consolidation; and 3) no chest radiograph report available but admission radiograph documented in the clinical notes as being in keeping with pneumonia (n=24). Clinical observations were matched to criteria of the National Health Service Swine Flu Community Assessment Tool for adults and children aged <16 years (online supplementary table A) .
FLU-CIN was a public health surveillance project for which the Ethics and Confidentiality Committee of the National Information Governance Board for Health and Social Care in England reviewed procedures and approved the collection, storage and use of personal data for surveillance purposes without the need for individual participant consent. This report represents a secondary analysis of the entire FLU-CIN database that includes data on 1520 patients from both spring/summer 2009 (n=601) and autumn/winter 2009/2010 (n=919) pandemic waves.
Descriptive analyses were performed comparing patients with and without asthma. The diagnosis of asthma was based on records in the medical case notes at the time of hospital admission and did not require documentation of confirmatory spirometry. These analyses were also performed separately for children (aged <16 years) and adults; p-values were calculated using the Chi-squared test for categorical exposure variables.
A nested case–control analysis of the surveillance cohort was used to investigate the association between a diagnosis of asthma and severe outcome from H1N1pdm 2009 infection, the latter being defined as death or the need for level 2 (high dependency unit) or level 3 (intensive care unit) care while in hospital. Patients with “severe outcomes” constituted “cases”. Potential confounding variables in the multivariable logistic regression were identified from both our previous work investigating risk factors for severe outcomes in the same cohort  and an a priori conceptual framework based on the descriptive analyses. The effect of the following covariates on the relationship between asthma and severe outcomes was independently assessed by introducing them separately, in turn, into the original model: age; comorbidities (using the Charlson index, excluding asthma cases) ; immune compromise; season of admission; severity of illness at admission (severe respiratory distress; increased respiratory rate; oxygen saturation ≤92%; respiratory exhaustion; severe clinical dehydration or shock; altered consciousness; dyspnoea and C-reactive protein levels); pre-admission inhaled steroid use; time to hospital admission (from symptom onset); in-hospital antiviral use; in-hospital antibiotic use; and in-hospital oral/intravenous corticosteroid use. Any covariate that modified the original unadjusted odds ratio by at least 10% towards a null association was then included in the final multivariable logistic regression model. If the association of asthma with less severe outcomes was fully explained by associated comorbidities or treatments, then the adjusted odds ratio of asthma with severe outcomes would be expected to approximate one in the final multivariate model.
By way of a further check on the robustness of our findings, an expanded multivariable analysis was also performed including all covariates that were independently associated with both exposure (asthma) and outcome. Results are presented as adjusted odds ratios and 95% confidence intervals. A p-value of <0.05 was taken to be statistically significant. STATA (release 11; StataCorp, College Station, TX, USA) was used for all statistical analyses.
The study cohort comprised 1520 patients; 720 (47.4%) were male and of median age 26 years (interquartile range 9–44.4), of whom 480 (31.6%) were children aged <16 years. Asthma was the commonest comorbid illness, present in 385 (25.3%) patients. Patients with asthma were more likely to be on regular inhaled corticosteroids than patients without asthma (52.7% versus 4.5%).
Tables 1 and and22 compare the demographic, clinical and treatment/management characteristics of asthmatic and nonasthmatic patients. Patients with asthma were significantly less likely to die or require intensive care support while in hospital compared with nonasthmatics (11.2% versus 19.8%; OR 0.51, 95% CI 0.36–0.72). Conversely, asthmatics were significantly more likely than nonasthmatics to exhibit features of severe respiratory compromise at the time of hospital admission, meeting the following triage criteria: severe respiratory distress, oxygen saturation ≤92%, respiratory exhaustion and severe clinical dehydration or clinical shock (table 1). Consistent with these findings, asthmatics reported significantly more dyspnoea at the time of admission (58.4% versus 30.8%; unadjusted OR 3.15, 95% CI 2.48–4.0) and were significantly more likely to receive supplemental oxygen (36.4% versus 26%; unadjusted OR 1.63, 95% CI 1.27–2.08). These differences were similar when children and adults were analysed separately (online supplementary tables B and C).
There was no significant difference in the proportions of asthmatic patients with pneumonia compared with nonasthmatics (asthma 17.1% versus 16.6%; OR 1.04, 95% CI 0.77–1.42). As asthmatics were more likely to have a chest radiograph performed on admission (OR 1.57, 95% CI 1.17–2.11; p=0.003), sensitivity analysis investigating the association between pneumonia and asthma including only patients with a definite chest radiograph record was performed; no significant difference between asthmatics and nonasthmatics was found (OR 0.85, 95% CI 0.62–1.18; p=0.329).
Antivirals (319 (82.9%) out of 385 cases using oseltamivir and nine (2.3%) out of 385 cases using zanamivir), antibiotics and systemic corticosteroids were prescribed upon hospital admission in significantly more patients with asthma compared with nonasthmatics (table 2).
Data relating to the time to admission from symptom onset were available for 1083 (71.3%) patients in the study cohort. Of these patients, those with asthma were significantly more likely to be admitted to hospital within 4 days of symptom onset compared with nonasthmatics (254 (65.2%) out of 385 versus 602 (53.0%) out of 1135; OR 1.46, 95% CI 1.04–2.07). Median (interquartile range) time from symptom onset to hospital admission was the same for asthmatics (2 (1–3) days) as for nonasthmatics (2 (1–4) days; p=0.28).
The association of asthma with severe outcomes (unadjusted OR 0.51, 95% CI 0.36–0.72) was unaffected by age (adjusted OR 0.49, 95% CI 0.35–0.70), presence of comorbidities as measured by the Charlson score (adjusted OR 0.51, 95% CI 0.36–0.72), immune compromise (adjusted OR 0.51, 95% CI 0.36–0.72), in-hospital antiviral use (adjusted OR 0.49, 95% 0.34–0.69), in-hospital antibiotic use (adjusted OR 0.49, 95% CI 0.35–0.70) or season (spring versus autumn/winter and first versus second pandemic wave) of admission (adjusted OR 0.51, 95% CI 0.36–0.72). In a further analysis to avoid linear assumptions regarding age and based on differences in age between asthmatics and nonasthmatics (table 1), the association of asthma with severe outcomes was adjusted for age <5 years (adjusted OR 0.50, 95% CI 0.35–0.71) and age 16–45 years (adjusted OR 0.52, 95% CI 0.36–0.74).
In contrast, point estimates of the association of asthma with severe outcome were reduced by >10% when adjusted independently for pre-admission inhaled corticosteroid use (adjusted OR 0.63, 95% CI 0.42–0.94) and “admission ≤4 days from symptom onset” (adjusted OR 0.64, 95% CI 0.43–0.95) (further results are given in online supplementary table D). In a multivariate model including both these covariates (pre-admission inhaled corticosteroids and admission ≤4 days from symptom onset), the association of asthma with a decreased likelihood of severe outcomes then failed to maintain statistical significance (adjusted OR 0.73, 95% CI 0.45–1.17) (table 3). In the multivariate model including all covariates independently associated with both exposure (asthma) and outcome, the association of asthma with a decreased likelihood of severe outcomes was again nonsignificant (adjusted OR 0.62, 95% CI 0.36–1.05; p=0.075), but pre-admission inhaled corticosteroids appeared protective (adjusted OR 0.34, 95% CI 0.18–0.66; p=0.001) as was early hospital admission (adjusted OR 0.60, 95% CI 0.38–0.94; p=0.025) (table 4).
The impact of in-hospital systemic (oral/i.v.) corticosteroid use on the association of asthma with severe outcomes was complex. An interaction between asthma status and the association of systemic corticosteroids with severe outcomes was observed; in asthmatics, systemic corticosteroids were associated with a decreased likelihood of severe outcomes (OR 0.36 adjusted for pre-admission inhaled corticosteroids and admission ≤4 days from symptom onset, 95% CI 0.18–0.72; p=0.004), while in nonasthmatics, systemic corticosteroids were associated with an increased likelihood of severe outcomes (OR 3.53 adjusted for pre-admission inhaled corticosteroids and admission ≤4 days from symptom onset, 95% CI 2.16–5.78; p<0.001).
Post-hoc analysis of asthmatics on pre-admission inhaled corticosteroids compared with asthmatics not on pre-admission inhaled corticosteroids did not demonstrate any significant differences between the two groups in relation to the number of other (nonasthma) comorbid illnesses and presenting features, except that asthmatics on inhaled corticosteroids were significantly more likely to complain of dyspnoea (64.5% versus 51.7%, unadjusted OR 1.7, 95% CI 1.13–2.56) and to meet triage criteria for severe respiratory distress (73.9% versus 62.1%, unadjusted OR 1.73, 95% CI 1.12–2.67) (table 5). Asthmatics on inhaled corticosteroids were significantly less likely to suffer a severe outcome compared with asthmatics not on inhaled steroids (7.4% versus 15.4%, unadjusted OR 0.44, 95% CI 0.23–0.85). Of the 254 patients on pre-admission inhaled corticosteroids, 51 (20.1%) patients were designated nonasthmatics. There was no evidence of benefit from pre-admission inhaled corticosteroids in nonasthmatics (unadjusted OR 0.99, 95% CI 0.49–2.00; p=0.968).
Asthma was the commonest comorbid illness in the FLU-CIN cohort, as observed consistently in other national and international cohorts [5, 7, 8]. In keeping with previous studies [5, 7, 8], a striking reduction in the likelihood of dying or requiring intensive care level support in asthmatics compared with nonasthmatics was noted. However, we found no evidence to suggest that asthmatics had a milder illness at the time of hospital admission to explain the lower rates of intensive care unit admission and death. Instead, the proportions of patients with radiologically confirmed pneumonia were very similar in asthmatics and nonasthmatics (17% versus 16.6%). In addition, compared with nonasthmatics, asthmatics had features indicative of greater respiratory compromise at the time of hospital admission; the latter being reflected in higher rates of dyspnoea, severe respiratory distress, respiratory exhaustion, oxygen saturations ≤92% and need for supplemental oxygen. These features suggest that a lower threshold for hospital referral and admission in patients with asthma is an unlikely explanation for the observed association of asthma with less severe outcomes in hospitalised patients.
Interestingly, nonasthmatics were significantly more likely to be immunocompromised (n=40) compared with asthmatics (n=2). A Canadian cohort study comparing hospitalised solid organ transplant recipients and nonimmunocompromised patients with pandemic H1N1pdm infection also found that asthma was more common in nonimmunocompromised patients . The reasons behind these observations warrant future investigation.
Although patients with asthma were more likely to receive antiviral and antibiotic treatment following admission to hospital, the association of asthma with less severe clinical outcomes was not influenced by in-patient antiviral use, antibiotic use, or by differences in age. Instead, factors related to the management of asthma were found to be more relevant; specifically the use of pre-admission inhaled corticosteroids, earlier admission to hospital (within 4 days of symptom onset) and in-hospital systemic (oral or intravenous) corticosteroid use.
Limitations of this study were that diagnoses of asthma recorded in case notes were not verified by reference to baseline spirometric data; in addition, the severity of airflow obstruction at the time of hospital admission, as measured by peak expiratory flow rates, was not captured in the FLU-CIN database. However, the lack of peak flow data would not have affected the main analysis comparing asthmatics versus nonasthmatics. The effect of comorbidities on the association of asthma with severe outcomes was adjusted for using the Deyo adaptation of the Charlson index score, which is a weighted measure using International Classifiaction of Diseases-10 classification. However, no comorbidity score is perfect and residual confounding cannot be completely excluded in this observational dataset. Data on time from symptom onset to hospital admission were available in 1,083 (71.3%) patients. There was no systematic reason for missing data and patients with missing data were similar to other patients. Therefore, given the overall size of the FLU-CIN cohort and the strength of the associations examined, it is unlikely that the missing data resulted in significantly altered results. Pandemic influenza vaccination status was not reliably recorded in the medical notes for all patients in the FLU-CIN cohort. Asthmatics, being in the target group for vaccination in the UK, may have been more likely to have been vaccinated compared with nonasthmatics. As all patients in this study had laboratory-confirmed H1N1pdm infection, if any patient had received the pandemic influenza vaccination, it had not prevented occurrence of infection. The possibility that vaccination of asthmatics may have attenuated the severity of illness remains. However, patients with other comorbid illnesses (who would have been equally likely to receive early vaccination) were not observed to have fewer severe outcomes, as had asthmatics. Therefore, a protective effect from pandemic influenza vaccination is unlikely to explain the findings of the current study.
The value of early hospital referral of patients with exacerbations of asthma triggered by influenza infection may relate to the earlier administration of antivirals which, in large cohort and population-based studies, has been associated with improved outcomes . Alternatively, this may reflect the importance of prompt appropriate treatment of asthma, including the administration of systemic corticosteroids, particularly in asthmatics not already on corticosteroid treatment.
Another view is that inhaled steroid use and early hospital admission might simply be markers of well-treated and confirmed asthma managed by well-informed doctors; and that asthmatics may have a different sort of illness from some nonasthmatics, this illness tending not to have such a severe ultimate outcome. The veracity of this view could not be explored within the current dataset. Patients with other chronic respiratory disorders, such as chronic obstructive pulmonary disease, typically managed with inhaled and systemic steroids and were not observed to have a lower likelihood of severe outcomes. This may reflect differences in the degree of response to treatment according to severity and chronicity of underlying lung disease, with asthma generally being more responsive than other chronic lung diseases .
The association between inhaled corticosteroid use in asthmatics and improved clinical outcomes is supported by findings from large epidemiological studies. Using the Saskatchewan Health databases and a cohort of 30 569 asthmatics, Suissa et al.  demonstrated that regular use of low-dose inhaled steroids was associated with a decreased risk of death from asthma and with reductions of 31% in the rate of hospital admissions for asthma . Likewise, low adherence rates to asthma medication, particularly inhaled corticosteroids, are associated with poorer asthma outcomes [20, 21]. The benefits from inhaled corticosteroids are thought to relate to their effects on airway inflammation and hyperresponsiveness and are not specific to any particular viral infection. Thus, although the current study focused on H1N1pdm infection, the observed benefit from inhaled corticosteroids on clinical outcomes in asthmatics would probably extend to other acute respiratory viral infections.
The relationship between asthma, in-hospital systemic corticosteroid use and clinical outcomes is more complicated. Some cohort studies of critically ill patients with H1N1pdm infection have reported an association of systemic corticosteroids with higher mortality or higher rates of hospital-acquired pneumonia [22–24], while other studies have not identified any association between systemic corticosteroid use and severe outcomes . None of these studies differentiated between asthmatics and nonasthmatics. In the current study, systemic corticosteroids were associated with a lower likelihood of severe outcomes in asthmatics, while in nonasthmatics, there was an association with an increased likelihood of severe outcomes. This reflects the substantially different indications for administration of systemic corticosteroids in asthmatics compared with nonasthmatics. The benefit of systemic corticosteroids in the acute management of exacerbations of asthma is well described and rests on a large evidence base . The findings from this study support the principles of acute asthma management, including the use of systemic corticosteroids for asthmatics hospitalised with influenza infection. In contradistinction, for nonasthmatics, the role of systemic corticosteroids in the management of severe influenza infections remains unclear and may even be harmful [27, 28]. In this study, corticosteroid use in nonasthmatics may have been more frequent in severely ill cases and, therefore, its use may merely be a marker for severe disease. However, this explanation alone does not account for the finding by Han et al.  that patients with H1N1pdm infection who were treated prospectively with corticosteroids as a fever suppressant suffered worse outcomes. An adequately powered randomised controlled trial is required to inform the current debate, taking into account the possibility that different clinical groups may have different types of disease and may either benefit from or be harmed by steroid therapy.
Validation of our results in a separate cohort would be a desirable next step. In the meantime, the findings from this study emphasise the importance of compliance with regular inhaled corticosteroid use by asthmatics, especially in winter when respiratory viruses are circulating. This study also supports the use of systemic corticosteroids for asthmatics hospitalised with influenza infection, in accordance with the principles of acute asthma management .
In conclusion, we found that asthmatics admitted to hospital with influenza infection were half as likely as nonasthmatics to die or require intensive care support despite greater respiratory compromise at the time of hospital admission and similar rates of pneumonia. Pre-admission use of inhaled corticosteroids and earlier admission to hospital (within 4 days) explained most of the observed association of asthma with less severe outcomes. This is consistent with the interpretation that a proportion of asthmatics, compared with nonasthmatics, were admitted because of a condition triggered by viral infection (i.e. virus-induced asthma exacerbation) rather than because of the severity of the infection itself and for which hospital treatment was more effective.
The authors thank E. Gadd and C. Armstrong (Department of Health, London, UK) for their invaluable contribution in relation to the FLU-CIN programme. They also gratefully acknowledge individuals who helped identify cases and collated clinical data: A. Booth, M. Charlesworth, S. Rodenhurst, A. Ballard and A. Holmes (Imperial College Healthcare NHS Trust, London, UK); S. Batham, P. Parkinson, T. Kumar and A. Dunphy (University Hospitals of Leicester NHS Trust, Leicester, UK); A. Tunbridge, P. Hempsall, J. Linskill, A. Turner, S. Grindle, D. Shelvin and E. Moulds (Sheffield University Hospitals NHS Trust, Sheffield, UK); E. White, E. Scott, J. Cater, E. Sergi and H. Hill (Alder Hey Children's Hospital NHS Foundation Trust, Liverpool, UK); D. Fleetwood, L. Roche, S. Dyas and M. Boswell (Royal Liverpool and Broadgreen University Hospital's Trust, Liverpool, UK); G. Vernon, G. Houghton, H. Longworth and A. Kerrigan (Liverpool Women’s Hospital, Liverpool, UK); G. Thompson, E. Jarvis and C. Minter (Nottingham University Hospitals NHS Trust, Nottingham, UK); K. Lum Kin, J. Daglish, S. Hayton and G. Slinn (Birmingham Children's Hospital, Birmingham, UK); M. Lacey, A. McAllister, K. Duffy, A. Gordon, K. Rooney, E. Anderson, M. McWhirter, H. Davison, W. Carmen, A. MacConnachie, M. Cotton, A. Reynolds, H. Murdoch, K. Voy, R. Hague, and colleagues in Scotland who contributed to the FLU-CIN in Scotland; B. Smyth and C. Kearns (National Public Health Agency, Belfast, UK) for identifying cases and facilitating data collection; T. Cunningham (Southern Trust, Portadown, UK) and L. Boydell (Belfast Trust, Belfast, UK) for facilitating data collection; A. Amberbir, S. Al-Badri, B. Mahgoob and N. Arunachalam (University of Nottingham, Nottingham, UK) for data entry and obtaining background population data; also G. Watson (University of Nottingham, Nottignham, UK) for database development and support. The authors also thank Professor Sir G. Duff, Co-Chair of the Scientific Advisory Group for Emergencies, and Professor J. Darbyshire, Co-Chair of the Influenza Clinical Information Network Strategy Group, for their support; S. Kelly (University of Nottingham) for help in developing the initial FLU-CIN data collection tool; and P. O'Brien (University College London Hospitals NHS Foundation Trust) for assistance in further developing the data set. The authors also thank those Chief Executive Officers, clinicians, virologists and managers too numerous to mention who were active in notifying cases to FLU-CIN. S.J. Brett and P.J. Openshaw wish to acknowledge the support of the UK NIHR Biomedical Research Centre scheme.
For editorial comments see page 772.
This article has supplementary material available from www.erj.ersjournals.com
The Department of Health in England was the main funder for FLU-CIN with additional funding from the Scottish Government Chief Medical Officer and Public Health Directorate for the centre in Scotland. Chairmanship of the FLU-CIN Strategy Group (that includes J.S. Nguyen-Van-Tam, W.S. Lim, M.G. Semple, P.J. Openshaw, R.C. Read, B.L. Taylor, S.J. Brett, J. McMenamin, J.E. Enstone and K.G. Nicholson as members) was provided by the Department of Health, England. The FLU-CIN Strategy Group had final responsibility for the interpretation of findings and decision to submit for publication.
Statement of Interest
Conflict of interest information can be found alongside the online version of this article at www.erj.ersjournals.com