- Email: [email protected]
Nosocomial vs community-acquired pandemic influenza A (H1N1) 2009: a nested case–control study
Journal of Hospital Infection 82 (2012) 94e100 Available online at www.sciencedirect.com
Journal of Hospital Infection journal homepage: www.elsevierhealth.com/journals/jhin
Nosocomial vs community-acquired pandemic influenza A (H1N1) 2009: a nested caseecontrol study G. Khandaker a, H. Rashid a, *, Y. Zurynski b, P.C. Richmond c, J. Buttery d, H. Marshall e, f, M. Gold f, T. Walls g, B. Whitehead h, E.J. Elliott b, i, R. Booy a a
National Centre for Immunisation Research and Surveillance of Vaccine Preventable Diseases, The Children’s Hospital at Westmead and The University of Sydney, Sydney, New South Wales, Australia b Discipline of Paediatrics and Child Health, Sydney Medical School, The University of Sydney, Sydney, and the Australian Paediatric Surveillance Unit, New South Wales, Australia c School of Paediatrics and Child Health, University of Western Australia, Perth, Western Australia, Australia d Murdoch Children’s Research Institute, Royal Children’s Hospital, Department of Paediatrics, Monash University, Melbourne, Victoria, Australia e Vaccinology and Immunology Research Trials Unit, Women’s and Children’s Hospital, Women’s and Children’s Health Network, South Australia, Australia f Discipline of Paediatrics, School of Reproductive Medicine and Paediatrics, University of Adelaide, South Australia, Australia g Paediatric Infectious Diseases, Sydney Children’s Hospital, Randwick, New South Wales, Australia h Respiratory Medicine, John Hunter Children’s Hospital, New South Wales, Australia i The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
A R T I C L E
I N F O
Article history: Received 10 February 2012 Accepted 2 July 2012 Available online 1 September 2012 Keywords: Australia Children Community-acquired influenza Nosocomial influenza Pandemic influenza
S U M M A R Y
Background: The characteristics of nosocomial influenza in children are not well described. Aim: To compare the characteristics of nosocomial and community-acquired pandemic influenza A (H1N1) 2009 (pH1N1) in Australian children. Methods: In a nested caseecontrol study, the clinical and epidemiological features of nosocomial vs community-acquired pH1N1 were compared among hospitalized children aged <15 years in six paediatric hospitals in Australia between 1 June and 30 September 2009. Findings: Of 506 hospitalized children with pH1N1, 47 (9.3%) were of nosocomial origin. These 47 cases were compared with 141 gender- and age-matched controls. Cases had a significantly higher proportion of underlying medical conditions compared with controls (81% vs 42%, P < 0.001), and were more likely to be exposed to household smokers (36% vs 20%, P ¼ 0.02). Fewer children with nosocomial influenza presented with classical symptoms of influenza, including subjective fever and lethargy. A higher proportion of children with nosocomial influenza received treatment with oseltamivir (77% vs 43%, P < 0.001), and they required a longer stay in hospital following the onset of influenza (mean 8.5 days vs 4.5 days, P ¼ 0.006). Three children (2%) in the community-acquired group died of pH1N1, but there were no deaths in the nosocomial group.
* Corresponding author. Address: National Centre for Immunisation Research and Surveillance of Vaccine Preventable Diseases, The Children’s Hospital at Westmead, Cnr Hawkesbury Rd and Hainsworth St, Locked Bag 4001, Westmead, NSW 2145, Australia. Tel.: þ61 2 9845 1489; fax: þ61 2 9845 1418. E-mail address: [email protected] (H. Rashid). 0195-6701/$ e see front matter ª 2012 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jhin.2012.07.006
G. Khandaker et al. / Journal of Hospital Infection 82 (2012) 94e100
95
Conclusion: This study shows that children with pre-existing diseases and those who are exposed to household smokers are more susceptible to nosocomial pH1N1. They may have ‘occult presentation’ of influenza, but their course of illness is not markedly different from that of children with community-acquired influenza. ª 2012 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved.
Introduction Nosocomial or healthcare-associated influenza is documented in the medical literature,1e7 but to the authors’ knowledge, there are no studies reporting comparative data on clinical and epidemiological features of nosocomial vs community-acquired influenza. Paediatric nosocomial influenza leads to extended hospital stay, excess morbidity and mortality, and economic loss.1 The latter has been estimated conservatively to be approximately US $7500 per patient for each additional stay in hospital.8 In epidemic/pandemic situations, this can strain an already overburdened healthcare system. During the first wave of the pandemic influenza A (H1N1) 2009 (pH1N1), the rate of all-cause nosocomial infections doubled in some countries (e.g. Mexico) compared with previous years.9 Subsequently, there have been several case reports, case series and outbreak investigations on nosocomial pandemic influenza in children.10e13 A recent surveillance report from the UK described 15 cases of paediatric nosocomial influenza; to the authors’ knowledge, this is the largest published report of nosocomial pH1N1 to date.14 The literature suggests a variable presentation and outcome of hospital-acquired pH1N1 influenza. Most case series or outbreak investigations reported that paediatric nosocomial pH1N1 was generally mild, rarely needed aggressive medical care and had very low mortality.10e13 However, the recent UK surveillance report indicated that the course and outcome of hospital-acquired pH1N1 can be severe both in terms of mortality and the need for escalated medical care.14 The report suggested that more than half of the children needed intensive medical care, and one-fifth of all children with nosocomial pH1N1 died. In order to compare the clinical and epidemiological features, risk factors and outcomes of nosocomial and community-acquired pH1N1 in children, a nested caseecontrol study was conducted at six major paediatric hospitals in Australia. A nested caseecontrol design was chosen for this study in order to achieve a statistically efficient analysis with matching for potential confounders.
Methods Setting This study is part of an active hospital-based surveillance system called ‘PAEDS’ (Paediatric Active Enhanced Disease Surveillance), which is a collaborative project between the Australian Paediatric Surveillance Unit (APSU) and the National Centre for Immunisation Research and Surveillance (NCIRS) that involves six tertiary paediatric referral centres in Australia (Children’s Hospital at Westmead in Western Sydney, Sydney Children’s Hospital in Eastern Sydney, John Hunter Children’s Hospital in Newcastle, Royal Children’s Hospital in Melbourne, Women’s and Children’s Hospital in Adelaide and Princes
Margaret Hospital in Perth). In 2009, the New South Wales Department of Health approved immediate commencement of the study in the three New South Wales hospitals without prior ethics approval, under provisions of the Health Records and Information Privacy Act 2002 (http://www.lawlink.nsw.gov.au). Subsequently, approval was also obtained from the human research ethics committee at each of the participating hospitals.
Participants Some aspects of the baseline data from the surveillance work have been published elsewhere with methodological details.15 Briefly, all admitted children aged <15 years who had laboratory-confirmed influenza in the participating hospitals between 1 June and 30 September 2009 were included. Nurses employed by PAEDS proactively identified children presenting at participating hospitals with signs and symptoms of influenza. Laboratory confirmation required a positive nasopharyngeal aspirate or combined nasopharyngeal and throat swabs for immunofluorescence; many children also had a rapid antigen test for influenza (A and B). Specimens positive for influenza A were referred for further typing by polymerase chain reaction (PCR) to state reference laboratories. A positive diagnosis was made when PCR was positive for pH1N1 influenza. Nosocomial influenza was considered if clinical signs and symptoms of influenza developed after 72 h of hospital admission for another reason. In case of transfer from other hospitals, a diagnosis of nosocomial influenza was made if the patient developed symptoms and signs of influenza after 72 h of admission in the referring hospital. Data on age, gender, symptoms and signs of the illness, preexisting chronic medical conditions and vaccination history, diagnostic test results, clinical course, complications and history of contact with a suspected case of pH1N1 were collected from relevant sources: guardians, patients’ hospital records and the Australian Childhood Immunization Register. Chronic medical conditions were defined as those illnesses that are recognized to increase the risk of influenza complications as listed in the Australian Immunisation Handbook (9th edition).16 Bacterial co-infections were defined as isolation of a bacterial pathogen from blood. A nested caseecontrol comparison was performed: children with nosocomial pH1N1 influenza were defined as ‘cases’, and children with community-acquired pH1N1 who were hospitalized during the study period were defined as ‘controls’. For every case with nosocomial pH1N1 infection, three controls were selected that were matched by gender and age. The controls were not necessarily from the same hospital as the case patient.
Statistical analysis Data analysis was performed using Statistical Package for the Social Sciences Version 19 (IBM Corporation, New York, NY,
96
G. Khandaker et al. / Journal of Hospital Infection 82 (2012) 94e100
USA). Initial comparison was made between the groups on several sets of variables (e.g. demographics, pre-existing medical conditions, clinical features, and course and outcome of the illness) using Chi-squared test or Fisher’s exact test where appropriate, and the mean differences between groups were compared using Student’s t-test or Wilcoxon ranksum test as applicable. For matched analysis, conditional logistic regression was used to estimate odds ratios (OR) and 95% confidence intervals (95% CI). Independent variables that were found to be associated with nosocomial influenza on univariate analysis with a significance level of P < 0.2 were considered for inclusion in the multi-variate model, except for composite variables such as ‘any underlying medical conditions’ and ‘absence of cough, coryza and fever triad’. A P-value <0.05 was considered to indicate significance.
Results Demographics There were 506 children with laboratory-confirmed pH1N1 in the PAEDS database, and 47 (9.3%) were cases of nosocomial pH1N1 influenza. Of the 459 (90.7%) children with communityacquired pH1N1 influenza, 141 (three per case) were selected as controls for the nested caseecontrol study. The distribution of cases and controls across the hospitals ranged from two to 22 and from nine to 42 subjects per hospital, respectively. The demographic characteristics of the two groups are shown in Table I.
Table II Comparison of underlying disorders in cases and controls: univariate and multi-variate analyses Risk factors Households with a smoker Any underlying medical condition Mean number of underlying conditions Chronic neurological conditions Immunosuppression Chronic heart diseases Chronic metabolic disorders Chronic liver diseases Bronchial asthma Chronic renal diseases Diabetes Other lung diseases (except bronchial asthma)
Cases N (%)
Controls P-value N (%)
17 (36.2) 28 (19.9) 0.02 38 (80.9) 59 (41.8) <0.001 1.47 0.64 0.03 14 11 7 6 4 4 2 1 0
Multi-variate analysis Households with a smoker Chronic neurological conditions Immunosuppression Chronic heart diseases Chronic metabolic disorders Chronic liver diseases Other lung diseases (except asthma)
(29.8) (23.4) (14.9) (12.7) (8.5) (8.5) (4.2) (2.1) (0)
10 14 3 3 2 17 3 4 10
(7.1) <0.001 (9.9) 0.02 (2.1) 0.001 (2.1) 0.003 (1.4) 0.02 (12.1) 0.50 (2.1) 0.43 (2.8) 0.79 (7.1) 0.06
Matched OR (95% CI) 3.3 10.4 4.4 23.5 2.6 3.3 0.03
(1.1e9.6) (2.6e40.5) (1.2e16.4) (1.9e293.5) (0.4e2.6) (0.2e57.5) (0.01e14.1)
P-value 0.03 0.001 0.03 0.01 0.34 0.42 0.98
OR, odds ratio; CI, confidence interval.
Risk factors Risk factors and comorbidities are shown in Table II. On univariate analysis, chronic neurological conditions, immunosuppression and chronic heart/metabolic/liver diseases were significantly higher in cases than in controls. However, on multi-variate analysis, chronic neurological, immunosuppressive and cardiac conditions remained significant. Compared with controls, cases were significantly more likely to be
Table I Demographic details of cases and controls Demographics
Cases
Controls
Median age in 45.6 46.8 months (range) (0.9e178.1) (0.7e173.5) Male:female 31:16 93:48 Mean body weight 18.1 (16.7) 17.8 (22.4) in kg (SD) Seasonal flu vaccination 3 (7.5)a 3 (2.5)a rate in 2009, N (%) Ethnicity, N (%) Caucasian 29 (61.7) 78 (55.3) Aboriginal and Torres 3 (6.4) 4 (2.8) Strait Islander Other 15 (31.9) 59 (41.8)
P-value 0.95 1.0 0.24 0.17
exposed to household smokers on both univariate and multivariate analyses.
Clinical presentation The clinical features of both groups are summarized in Table III. Although cough, coryza and fever were the most common presenting symptoms, 38% of children with nosocomial influenza did not have any of these symptoms, compared with 21% of children with community-acquired influenza (P ¼ 0.02). On univariate analysis, cough, subjective fever, lethargy and myalgia/arthralgia were significantly less common in children with nosocomial influenza than in children with community-acquired influenza. However, on multi-variate analysis, only subjective fever and lethargy remained significant. All other clinical symptoms of influenza tended to be less common in children with nosocomial influenza than in children with community-acquired influenza, but this difference was not significant.
Course and outcome 0.44 0.24 0.23
SD, standard deviation. a Forty children in the case group and 120 children in the control group were eligible for influenza vaccine.
The courses and outcomes of the illness have been summarized for both groups in Table IV. The most common complication of pH1N1 influenza was pneumonia, which affected just under 20% of children in each group. A higher proportion of children with nosocomial pH1N1 had bloodculture-proven bacterial co-infection during their course of
G. Khandaker et al. / Journal of Hospital Infection 82 (2012) 94e100 Table III Comparison of clinical features in cases and controls: univariate and multi-variate analyses Clinical features
Cases N (%)
Cough 23 Coryza 21 Fever (documented) 14 Fever (self-reported) 10 Dyspnoea/shortness 9 of breath Vomiting 8 Lethargy 6 Diarrhoea 5 Sore throat 2 Headache 2 Seizure 2 Myalgia/joint pain 0 Absence of ‘cough, 18 coryza and fever triad’ Mean highest recorded 15.3 C-reactive protein Multi-variate analysis Cough Fever (self-reported) Vomiting Lethargy Sore throat Myalgia/joint pain
Controls N (%)
P-value
(48.9) (44.7) (29.8) (21.3) (19.1)
98 69 53 70 36
(69.5) (48.9) (37.6) (49.6) (25.5)
0.01 0.61 0.33 <0.001 0.37
(17.0) (12.8) (10.6) (4.3) (4.3) (4.3) (0) (38.3)
40 55 20 18 13 11 12 28
(28.4) (39) (14.2) (12.8) (9.2) (7.8) (8.5) (20.6)
0.12 <0.001 0.54 0.10 0.28 0.41 0.04 0.02
(SD 65.0) 8.1 (SD 28.7)
Matched OR (95% CI) 1 0.3 0.8 0.3 0.3 0.03
(0.4e2.6) (0.1e0.8) (0.3e2.3) (0.1e0.9) (0.1e1.6) (0.01e8.2)
0.1
P-value 0.99 0.02 0.71 0.03 0.16 0.98
SD, standard deviation; OR, odds ratio; CI, confidence interval.
illness (6.4% vs 1.4%, P ¼ 0.09). Co-infection with other viruses was found in six (4.3%) patients in the community-acquired group (respiratory syncytial virus, adenovirus and parainfluenza virus), and one (2.1%) patient in the nosocomial group (respiratory syncytial virus). A higher proportion of children in the nosocomial group received treatment with oseltamivir (77% vs 43%, P < 0.001) and antibiotics (75% vs 55%, P ¼ 0.02) compared with children in the community-acquired group, although only treatment with oseltamivir remained significant on multi-variate analysis (matched OR 3, 95% CI 1.4e7.5). The use of antibiotics occurred during the course of influenza infection; however, the indication may or may not be related to influenza or its complications. Three-quarters of children in both groups received acetaminophen, and approximately 30% of children in both groups received non-steroidal anti-inflammatory drugs (NSAIDs). In the community-acquired group, three boys aged four, nine and 12 years died during the period of hospitalization. The four-year-old had a brainstem glioblastoma; the nine-year-old had spastic quadriplegia, developmental delay and epilepsy; and the 12-year-old had no known pre-existing medical condition but died one day after hospitalization. All three deaths were attributed to influenza. There were no hospital deaths in the nosocomial group. However, one child died three months after discharge from hospital. He was a two-month-old boy admitted with cholestatic jaundice and failure to thrive, and remained in hospital
97
Table IV Comparison of course of illness and outcomes in cases and controls: univariate and multi-variate analyses Course and outcome Length of hospital stay (days) Length of post-influenza hospital stay (days) ICU admission Length of stay in ICU (days) Ventilated Duration of ventilation (days) Treated with oseltamivir Treated with antibiotics Received NSAIDs Received acetaminophen Developed pneumonia Bacterial co-infectiona Other complicationsd
Cases N (%)
Controls N (%) P-value
13.4 (SD 12.2) 4.5 (SD 8.5) <0.001 8.5 (SD 14.5) 4.5 (SD 8.5)
0.006
7 (14.9) 9.1 (SD 8.6)
0.43 0.95
15 (10.6) 7.0 (SD 8.4)
5 (10.6) 11 (7.8) 3.2 (SD 1.7) 11.2 (SD 16.9)
Multi-variate analysis Treated with oseltamivir Treated with antibiotics
0.37 0.21
36 (76.6)
60 (42.6)
<0.001
35 (74.5)
77 (54.6)
0.02
15 (31.9) 38 (80.9)
39 (27.6) 104 (73.8)
0.58 0.63
8 (17.0) 3 (6.4)b 6 (12.8)
26 (18.4) 2 (1.4)c 15 (10.6)
0.83 0.09 0.67
Matched OR (95% CI)
P-value
3 (1.4e7.5) 1.4 (0.6e3.1)
0.006 0.48
ICU, intensive care unit; NSAIDs, non-steroidal anti-inflammatory drugs; SD, standard deviation; OR, odds ratio; CI, confidence interval. a Numbers are too small to be included in multi-variate model to obtain a meaningful result. b Identified bacteria were Escherichia coli, Acinetobacter spp. and Micrococcus spp. c Identified bacteria were Haemophilus influenzae and Streptococcus pneumoniae. d Pleural effusion, pneumothorax, acute respiratory distress syndrome, convulsion and encephalitis.
for approximately one month while he had pH1N1 infection. He died three months and eight days after discharge from hospital due to causes unrelated to pH1N1. Data on influenza contacts were available for 42 (of 47) children with nosocomial influenza and 105 (of 141) children with community-acquired influenza. Only five of 42 (12%) patients with nosocomial influenza had an identified contact: two children reported contact with other influenza-positive patients, two children reported contact with a symptomatic family member, and one child reported contact with an ill school friend. It was assumed that contact with hospital staff and/or visitors was important for the others. Thirty-six of 105 (34%) children with community-acquired influenza had an identified contact: the majority (83%) reported contact with symptomatic family members, and the rest reported contact with school friends with influenza-like illness.
Discussion This study has described the largest known cohort of children (N ¼ 47) with nosocomial influenza in a single season, and
98
G. Khandaker et al. / Journal of Hospital Infection 82 (2012) 94e100
the first nested caseecontrol study on nosocomial vs community-acquired pH1N1. The only other large cohort of hospital-acquired influenza has been described from Philadelphia Children’s Hospital, involving 46 children aged 21 years hospitalized over four influenza seasons from 2000 onwards.6 This study found that nosocomial influenza differed significantly from community-acquired influenza in terms of disease susceptibility and severity, clinical presentation and aspects of the course of illness. Eighty-one percent of children with nosocomial influenza had an underlying chronic medical condition, compared with 42% of children with communityacquired influenza. Multi-variate analysis showed that chronic neurological, immunosuppressive and cardiac conditions were independent predictors of nosocomial infection, but chronic metabolic, hepatic and pulmonary diseases other than asthma lost significance (Table II). Additionally, children with nosocomial pH1N1 had, on average, a greater number of underlying medical illnesses than children with communityacquired pH1N1 (1.5 vs 0.6, P ¼ 0.03). This is consistent with other studies. Enstone et al. reported that 93% of UK children with nosocomial pH1N1 had serious underlying illnesses, while Fanella et al. from Canada reported that all nosocomial cases had underlying medical conditions.13,14 Both host and environmental factors could be responsible. Generalized immunosuppression, poor functioning of a specific organ/system (e.g. respiratory or cardiac disease) or a combination of both may render these children more susceptible to influenza. Environmentally, as these children require frequent hospitalization for extended periods and need multi-disciplinary care by multiple healthcare workers (HCWs), they may be exposed regularly to patients, visitors or HCWs with influenza. Chronic neurological, immunosuppressive, cardiac and metabolic conditions were the most common pre-existing medical conditions identified in this study, and this is comparable with findings from other studies.6,14 This study also found that children with nosocomial pH1N1 were significantly more likely to have a smoker in the household compared with children with community-acquired disease. Active smoking is known to increase the risk of influenza,17e20 and passive smoking is known to increase the risk of contracting lower respiratory tract infections requiring hospitalization in early life.21,22 However, the mechanism for the increased risk of influenza due to passive smoking or exposure to smokers has not been described in detail to date. Experimental studies suggested that smoke exposure impairs influenza-induced mucosal innate immunity (e.g. through suppression of interleukin-6).18,23 It is also likely that parents who smoke, who tend to cough more than parents who do not smoke, may spread the droplets containing influenza virus through coughing during their visits to their hospitalized children. In this study, children with nosocomial influenza had significantly fewer classical symptoms of pH1N1, particularly subjective fever and lethargy, compared with children with community-acquired influenza.24 Eighty percent (12/15) of UK children with nosocomial pH1N1 presented with fever, but other symptoms were not so common.14 Immunosuppression and concurrent use of antipyretics, analgesics and other medications may modify the classical presentation of influenza in children hospitalized with pre-existing chronic conditions. Babcock et al. commented that patients acquiring influenza while hospitalized were less likely to present with typical
symptoms of influenza, possibly because of concurrent medications with anti-inflammatory and antipyretic effects.25 In the present study, 81% of children with nosocomial influenza received acetaminophen and 32% received NSAIDs during the course of their hospitalization. Interestingly, a similar proportion of children with community-acquired influenza also received these drugs, so the difference cannot be explained purely by the use of anti-inflammatory or antipyretic medications. It may be a special feature of nosocomial influenza to have ‘occult’ or ‘mild’ presentation. Goldwater et al. reported that a respiratory virus (including influenza) or mycoplasma was detected in 46% of children with nosocomial respiratory infection in whom symptoms were initially overlooked, and a respiratory virus or mycoplasma was detected in another 47% of asymptomatic children.5 Influenza is generally a mild illness and only a small proportion of community-acquired disease is sufficiently severe to require hospitalization. As children with nosocomial influenza were inpatients for other reasons, heightened awareness during the pandemic season may have allowed easy detection of milder cases. The UK surveillance study may have underestimated the actual burden of nosocomial influenza. It reported that only 2% of cases of hospitalized children with pH1N1 met the criteria of nosocomial infection.14 In a retrospective analysis of five years of data from a Canadian paediatric hospital, Slinger and Dennis demonstrated that nosocomial influenza accounted for 7.6% of all hospitalized cases of influenza, which is closer to the present finding of 9.3%.4 Therefore, it is possible that the UK surveillance study only identified the more severe pH1N1 cases (there were high rates of death or admission to intensive care), and missed the asymptomatic and less symptomatic cases, resulting in an overestimated case fatality ratio of 20%.14 Such a high case fatality ratio was not seen in other studies.10e13 The present authors believe that if well-designed prospective surveillance is in place, more cases will be found, many of which may be intrinsically mild; if such patients have already been discharged, they would not be ill enough to need to return to hospital. A high index of suspicion is needed to identify influenza in hospitalized children with underlying risk factors to prevent transmission and enable early treatment. More children with nosocomial pH1N1 had blood-cultureproven bacterial co-infection compared with children with community-acquired pH1N1 (6.4% vs 1.4%, P ¼ 0.09). Otherwise, complications in both groups were similar, with just under 20% of children in each group developing pneumonia. This is approximately half of what has been reported from the UK.14 The higher proportion of antibiotic use and insignificantly higher levels of C-reactive protein found in the children with nosocomial influenza may be due to the higher proportion of bacterial co-infection in these children compared with the children with community-acquired influenza. This is supported by a study from Western Australia.26 As expected, the mean duration of hospitalization was longer in the children with nosocomial influenza (13.4 vs 4.5, P < 0.001), as was the length of post-influenza hospital stay (8.5 vs 4.5, P ¼ 0.006). Otherwise, the course of illness was essentially similar in both groups. Understandably, the use of antiviral therapy within two days of onset of illness was higher in children with nosocomial pH1N1 than in children with community-acquired pH1N1 (matched OR 3, 95% CI 1.4e7.5) given their other illnesses. It is not clear if prompt antiviral
G. Khandaker et al. / Journal of Hospital Infection 82 (2012) 94e100 therapy decreased the severity of illnesses in children with nosocomial influenza, but antiviral drug usage in this cohort was comparable with published data.13,14 A limitation of this study is that contact tracing was not performed actively. Only five of 42 (12%) patients with nosocomial pH1N1 whose data were available knew of a contact. Fanella et al. reported that almost all patients with nosocomial pH1N1 acquired the illness from another patient on the ward in their hospital in Canada.13 It is notable that HCWs are also important transmitters of the infection.1,7 Molecular genotyping can help describe the outbreaks of nosocomial pH1N1, although its wider applicability requires further assessment.27 Finally, it is important that all available preventive measures, whether pharmaceutical or non-pharmaceutical, are used to prevent nosocomial influenza in children with atrisk medical conditions. These include annual influenza vaccination of eligible children (children aged six months or more with a chronic medical condition predisposing to severe influenza illness that requires regular medical follow-up or hospitalization), vaccination of HCWs, use of personal protective measures, hand hygiene and tight hospital infection control measures.7 A recent study demonstrated the role of neuraminidase inhibitors in preventing the spread of nosocomial influenza between patients.28 In summary, nearly one-tenth of hospitalized children with pH1N1 in Australia acquired the infection in healthcare settings. Children with underlying medical conditions were at greater risk of developing nosocomial pH1N1. These children may have an ‘occult presentation’ of influenza, and therefore may be considerably under-reported. It is suggested that nosocomial influenza may be far more common than clinicians realize, and should be considered particularly during the influenza season. Conflict of interest statement RB has received financial support from CSL, Sanofi, GlaxoSmithKline, Novartis, Roche and Wyeth to conduct research and attend and present at scientific meetings. Any funding received is directed to an NCIRS research account at The Children’s Hospital at Westmead and is not personally accepted by RB. HM has participated as a board member of global advisory boards for Merck and GlaxoSmithKline, and has participated as an investigator in industry-sponsored clinical vaccine trials. JB has served on advisory and/or data safety monitoring committees for CSL Vaccines and GlaxoSmithKline, for which the Murdoch Children’s Research Institute receives payments into an educational fund. The other authors declare that they have no conflict of interest in relation to this work. Funding sources This project was funded, in part, by the New South Wales Department of Health and the National Health and Medical Research Council (NHMRC) H1N1 Grant No: 633028. Activities of the APSU and NCIRS are supported by the Australian Government Department of Health and Ageing. APSU is supported by the NHMRC (Enabling Grant No: 402784 and Practitioner Fellowship No: 457084, E. Elliott), NHMRC Career Development Fellowship (No. 1016272 H. Marshall), the Discipline of Paediatrics
99
and Child Health and Faculty of Medicine, University of Sydney; the Children’s Hospital at Westmead; and the Royal Australasian College of Physicians.
References 1. Maltezou HC, Drancourt M. Nosocomial influenza in children. J Hosp Infect 2003;55:83e91. 2. Hall CB, Douglas Jr RG. Nosocomial influenza infection as a cause of intercurrent fevers in infants. Pediatrics 1975;55: 673e677. 3. Cunney RJ, Bialachowski A, Thornley D, Smaill FM, Pennie RA. An outbreak of influenza A in a neonatal intensive care unit. Infect Control Hosp Epidemiol 2000;21:449e454. 4. Slinger R, Dennis P. Nosocomial influenza at a Canadian pediatric hospital from 1995 to 1999: opportunities for prevention. Infect Control Hosp Epidemiol 2002;23:627e629. 5. Goldwater PN, Martin AJ, Ryan B, et al. A survey of nosocomial respiratory viral infections in a children’s hospital: occult respiratory infection in patients admitted during an epidemic season. Infect Control Hosp Epidemiol 1991;12:231e238. 6. Leckerman KH, Sherman E, Knorr J, Zaoutis TE, Coffin SE. Risk factors for healthcare-associated, laboratory-confirmed influenza in hospitalized pediatric patients: a caseecontrol study. Infect Control Hosp Epidemiol 2010;31:421e424. 7. Salgado CD, Farr BM, Hall KK, Hayden FG. Influenza in the acute hospital setting. Lancet Infect Dis 2002;2:145e155. 8. Serwint JR, Miller RM. Why diagnose influenza infections in hospitalized pediatric patients? Pediatr Infect Dis J 1993;12: 200e204. 9. Perez-Padilla R, Fernandez R, Garcia-Sancho C, Franco-Marina F, Mondragon E, Volkow P. Demand for care and nosocomial infection rate during the first influenza AH1N1 2009 virus outbreak at a referral hospital in Mexico City. Salud Publica Mex 2011;53: 334e340. 10. Cunha BA, Thekkel V, Krilov L. Nosocomial swine influenza (H1N1) pneumonia: lessons learned from an illustrative case. J Hosp Infect 2010;74:278e281. 11. Redelman-Sidi G, Sepkowitz KA, Huang CK, et al. 2009 H1N1 influenza infection in cancer patients and hematopoietic stem cell transplant recipients. J Infect 2010;60:257e263. 12. Chironna M, Tafuri S, Santoro N, Prato R, Quarto M, Germinario CA. A nosocomial outbreak of 2009 pandemic influenza A(H1N1) in a paediatric oncology ward in Italy, Octobere November 2009. Euro Surveill 2010;15:19454. 13. Fanella ST, Pinto MA, Bridger NA, et al. Pandemic (H1N1) 2009 influenza in hospitalized children in Manitoba: nosocomial transmission and lessons learned from the first wave. Infect Control Hosp Epidemiol 2011;32:435e443. 14. Enstone JE, Myles PR, Openshaw PJ, et al. Nosocomial pandemic (H1N1) 2009, United Kingdom, 2009e2010. Emerg Infect Dis 2011;17:592e598. 15. Elliott EJ, Zurynski YA, Walls T, Whitehead B, Gilmour R, Booy R. Novel inpatient surveillance in tertiary paediatric hospitals in New South Wales illustrates impact of first-wave pandemic influenza A H1N1 (2009) and informs future health service planning. J Paediatr Child Health 2011;23:1440e1754. 16. National Health and Medical Research Council. The Australian immunisation handbook. 9th ed. Canberra: Australian Government Department of Health and Ageing; 2008. p. 184e194. 17. Ward KA, Spokes PJ, McAnulty JM. Caseecontrol study of risk factors for hospitalization caused by pandemic (H1N1) 2009. Emerg Infect Dis 2011;17:1409e1416. 18. Noah TL, Zhou H, Monaco J, Horvath K, Herbst M, Jaspers I. Tobacco smoke exposure and altered nasal responses to live attenuated influenza virus. Environ Health Perspect 2011;119: 78e83.
100
G. Khandaker et al. / Journal of Hospital Infection 82 (2012) 94e100
19. Arcavi L, Benowitz NL. Cigarette smoking and infection. Arch Intern Med 2004;164:2206e2216. 20. Muscatello DJ, Barr M, Thackway SV, Macintyre CR. Epidemiology of influenza-like illness during pandemic (H1N1) 2009, New South Wales. Australia. Emerg Infect Dis 2011;17:1240e1247. 21. Huttunen R, Heikkinen T, Syrjanen J. Smoking and the outcome of infection. J Intern Med 2011;269:258e269. 22. Strachan DP, Cook DG. Health effects of passive smoking. 1. Parental smoking and lower respiratory illness in infancy and early childhood. Thorax 1997;52:905e914. 23. Gualano RC, Hansen MJ, Vlahos R, et al. Cigarette smoke worsens lung inflammation and impairs resolution of influenza infection in mice. Respir Res 2008;9:53. 24. Khandaker G, Zurynski Y, Lester-Smith D, et al. Clinical features, oseltamivir treatment and outcome in infants aged <12 months
25.
26.
27.
28.
with laboratory-confirmed influenza A in 2009. Antivir Ther 2011;16:1005e1010. Babcock HM, Merz LR, Fraser VJ. Is influenza an influenza-like illness? Clinical presentation of influenza in hospitalized patients. Infect Control Hosp Epidemiol 2006;27:266e270. Ingram PR, Inglis T, Moxon D, Speers D. Procalcitonin and C-reactive protein in severe 2009 H1N1 influenza infection. Intensive Care Med 2010;36:528e532. Rodriguez-Sanchez B, Alonso M, Catalan P, et al. Genotyping of a nosocomial outbreak of pandemic influenza A/H1N1 2009. J Clin Virol 2011;1:1. Higa F, Tateyama M, Tomishima M, et al. Role of neuraminidase inhibitor chemoprophylaxis in controlling nosocomial influenza: an observational study. Influenza Other Respir Viruses 2012;6: 299e303.
Journal of Hospital Infection journal homepage: www.elsevierhealth.com/journals/jhin
Nosocomial vs community-acquired pandemic influenza A (H1N1) 2009: a nested caseecontrol study G. Khandaker a, H. Rashid a, *, Y. Zurynski b, P.C. Richmond c, J. Buttery d, H. Marshall e, f, M. Gold f, T. Walls g, B. Whitehead h, E.J. Elliott b, i, R. Booy a a
National Centre for Immunisation Research and Surveillance of Vaccine Preventable Diseases, The Children’s Hospital at Westmead and The University of Sydney, Sydney, New South Wales, Australia b Discipline of Paediatrics and Child Health, Sydney Medical School, The University of Sydney, Sydney, and the Australian Paediatric Surveillance Unit, New South Wales, Australia c School of Paediatrics and Child Health, University of Western Australia, Perth, Western Australia, Australia d Murdoch Children’s Research Institute, Royal Children’s Hospital, Department of Paediatrics, Monash University, Melbourne, Victoria, Australia e Vaccinology and Immunology Research Trials Unit, Women’s and Children’s Hospital, Women’s and Children’s Health Network, South Australia, Australia f Discipline of Paediatrics, School of Reproductive Medicine and Paediatrics, University of Adelaide, South Australia, Australia g Paediatric Infectious Diseases, Sydney Children’s Hospital, Randwick, New South Wales, Australia h Respiratory Medicine, John Hunter Children’s Hospital, New South Wales, Australia i The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
A R T I C L E
I N F O
Article history: Received 10 February 2012 Accepted 2 July 2012 Available online 1 September 2012 Keywords: Australia Children Community-acquired influenza Nosocomial influenza Pandemic influenza
S U M M A R Y
Background: The characteristics of nosocomial influenza in children are not well described. Aim: To compare the characteristics of nosocomial and community-acquired pandemic influenza A (H1N1) 2009 (pH1N1) in Australian children. Methods: In a nested caseecontrol study, the clinical and epidemiological features of nosocomial vs community-acquired pH1N1 were compared among hospitalized children aged <15 years in six paediatric hospitals in Australia between 1 June and 30 September 2009. Findings: Of 506 hospitalized children with pH1N1, 47 (9.3%) were of nosocomial origin. These 47 cases were compared with 141 gender- and age-matched controls. Cases had a significantly higher proportion of underlying medical conditions compared with controls (81% vs 42%, P < 0.001), and were more likely to be exposed to household smokers (36% vs 20%, P ¼ 0.02). Fewer children with nosocomial influenza presented with classical symptoms of influenza, including subjective fever and lethargy. A higher proportion of children with nosocomial influenza received treatment with oseltamivir (77% vs 43%, P < 0.001), and they required a longer stay in hospital following the onset of influenza (mean 8.5 days vs 4.5 days, P ¼ 0.006). Three children (2%) in the community-acquired group died of pH1N1, but there were no deaths in the nosocomial group.
* Corresponding author. Address: National Centre for Immunisation Research and Surveillance of Vaccine Preventable Diseases, The Children’s Hospital at Westmead, Cnr Hawkesbury Rd and Hainsworth St, Locked Bag 4001, Westmead, NSW 2145, Australia. Tel.: þ61 2 9845 1489; fax: þ61 2 9845 1418. E-mail address: [email protected] (H. Rashid). 0195-6701/$ e see front matter ª 2012 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jhin.2012.07.006
G. Khandaker et al. / Journal of Hospital Infection 82 (2012) 94e100
95
Conclusion: This study shows that children with pre-existing diseases and those who are exposed to household smokers are more susceptible to nosocomial pH1N1. They may have ‘occult presentation’ of influenza, but their course of illness is not markedly different from that of children with community-acquired influenza. ª 2012 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved.
Introduction Nosocomial or healthcare-associated influenza is documented in the medical literature,1e7 but to the authors’ knowledge, there are no studies reporting comparative data on clinical and epidemiological features of nosocomial vs community-acquired influenza. Paediatric nosocomial influenza leads to extended hospital stay, excess morbidity and mortality, and economic loss.1 The latter has been estimated conservatively to be approximately US $7500 per patient for each additional stay in hospital.8 In epidemic/pandemic situations, this can strain an already overburdened healthcare system. During the first wave of the pandemic influenza A (H1N1) 2009 (pH1N1), the rate of all-cause nosocomial infections doubled in some countries (e.g. Mexico) compared with previous years.9 Subsequently, there have been several case reports, case series and outbreak investigations on nosocomial pandemic influenza in children.10e13 A recent surveillance report from the UK described 15 cases of paediatric nosocomial influenza; to the authors’ knowledge, this is the largest published report of nosocomial pH1N1 to date.14 The literature suggests a variable presentation and outcome of hospital-acquired pH1N1 influenza. Most case series or outbreak investigations reported that paediatric nosocomial pH1N1 was generally mild, rarely needed aggressive medical care and had very low mortality.10e13 However, the recent UK surveillance report indicated that the course and outcome of hospital-acquired pH1N1 can be severe both in terms of mortality and the need for escalated medical care.14 The report suggested that more than half of the children needed intensive medical care, and one-fifth of all children with nosocomial pH1N1 died. In order to compare the clinical and epidemiological features, risk factors and outcomes of nosocomial and community-acquired pH1N1 in children, a nested caseecontrol study was conducted at six major paediatric hospitals in Australia. A nested caseecontrol design was chosen for this study in order to achieve a statistically efficient analysis with matching for potential confounders.
Methods Setting This study is part of an active hospital-based surveillance system called ‘PAEDS’ (Paediatric Active Enhanced Disease Surveillance), which is a collaborative project between the Australian Paediatric Surveillance Unit (APSU) and the National Centre for Immunisation Research and Surveillance (NCIRS) that involves six tertiary paediatric referral centres in Australia (Children’s Hospital at Westmead in Western Sydney, Sydney Children’s Hospital in Eastern Sydney, John Hunter Children’s Hospital in Newcastle, Royal Children’s Hospital in Melbourne, Women’s and Children’s Hospital in Adelaide and Princes
Margaret Hospital in Perth). In 2009, the New South Wales Department of Health approved immediate commencement of the study in the three New South Wales hospitals without prior ethics approval, under provisions of the Health Records and Information Privacy Act 2002 (http://www.lawlink.nsw.gov.au). Subsequently, approval was also obtained from the human research ethics committee at each of the participating hospitals.
Participants Some aspects of the baseline data from the surveillance work have been published elsewhere with methodological details.15 Briefly, all admitted children aged <15 years who had laboratory-confirmed influenza in the participating hospitals between 1 June and 30 September 2009 were included. Nurses employed by PAEDS proactively identified children presenting at participating hospitals with signs and symptoms of influenza. Laboratory confirmation required a positive nasopharyngeal aspirate or combined nasopharyngeal and throat swabs for immunofluorescence; many children also had a rapid antigen test for influenza (A and B). Specimens positive for influenza A were referred for further typing by polymerase chain reaction (PCR) to state reference laboratories. A positive diagnosis was made when PCR was positive for pH1N1 influenza. Nosocomial influenza was considered if clinical signs and symptoms of influenza developed after 72 h of hospital admission for another reason. In case of transfer from other hospitals, a diagnosis of nosocomial influenza was made if the patient developed symptoms and signs of influenza after 72 h of admission in the referring hospital. Data on age, gender, symptoms and signs of the illness, preexisting chronic medical conditions and vaccination history, diagnostic test results, clinical course, complications and history of contact with a suspected case of pH1N1 were collected from relevant sources: guardians, patients’ hospital records and the Australian Childhood Immunization Register. Chronic medical conditions were defined as those illnesses that are recognized to increase the risk of influenza complications as listed in the Australian Immunisation Handbook (9th edition).16 Bacterial co-infections were defined as isolation of a bacterial pathogen from blood. A nested caseecontrol comparison was performed: children with nosocomial pH1N1 influenza were defined as ‘cases’, and children with community-acquired pH1N1 who were hospitalized during the study period were defined as ‘controls’. For every case with nosocomial pH1N1 infection, three controls were selected that were matched by gender and age. The controls were not necessarily from the same hospital as the case patient.
Statistical analysis Data analysis was performed using Statistical Package for the Social Sciences Version 19 (IBM Corporation, New York, NY,
96
G. Khandaker et al. / Journal of Hospital Infection 82 (2012) 94e100
USA). Initial comparison was made between the groups on several sets of variables (e.g. demographics, pre-existing medical conditions, clinical features, and course and outcome of the illness) using Chi-squared test or Fisher’s exact test where appropriate, and the mean differences between groups were compared using Student’s t-test or Wilcoxon ranksum test as applicable. For matched analysis, conditional logistic regression was used to estimate odds ratios (OR) and 95% confidence intervals (95% CI). Independent variables that were found to be associated with nosocomial influenza on univariate analysis with a significance level of P < 0.2 were considered for inclusion in the multi-variate model, except for composite variables such as ‘any underlying medical conditions’ and ‘absence of cough, coryza and fever triad’. A P-value <0.05 was considered to indicate significance.
Results Demographics There were 506 children with laboratory-confirmed pH1N1 in the PAEDS database, and 47 (9.3%) were cases of nosocomial pH1N1 influenza. Of the 459 (90.7%) children with communityacquired pH1N1 influenza, 141 (three per case) were selected as controls for the nested caseecontrol study. The distribution of cases and controls across the hospitals ranged from two to 22 and from nine to 42 subjects per hospital, respectively. The demographic characteristics of the two groups are shown in Table I.
Table II Comparison of underlying disorders in cases and controls: univariate and multi-variate analyses Risk factors Households with a smoker Any underlying medical condition Mean number of underlying conditions Chronic neurological conditions Immunosuppression Chronic heart diseases Chronic metabolic disorders Chronic liver diseases Bronchial asthma Chronic renal diseases Diabetes Other lung diseases (except bronchial asthma)
Cases N (%)
Controls P-value N (%)
17 (36.2) 28 (19.9) 0.02 38 (80.9) 59 (41.8) <0.001 1.47 0.64 0.03 14 11 7 6 4 4 2 1 0
Multi-variate analysis Households with a smoker Chronic neurological conditions Immunosuppression Chronic heart diseases Chronic metabolic disorders Chronic liver diseases Other lung diseases (except asthma)
(29.8) (23.4) (14.9) (12.7) (8.5) (8.5) (4.2) (2.1) (0)
10 14 3 3 2 17 3 4 10
(7.1) <0.001 (9.9) 0.02 (2.1) 0.001 (2.1) 0.003 (1.4) 0.02 (12.1) 0.50 (2.1) 0.43 (2.8) 0.79 (7.1) 0.06
Matched OR (95% CI) 3.3 10.4 4.4 23.5 2.6 3.3 0.03
(1.1e9.6) (2.6e40.5) (1.2e16.4) (1.9e293.5) (0.4e2.6) (0.2e57.5) (0.01e14.1)
P-value 0.03 0.001 0.03 0.01 0.34 0.42 0.98
OR, odds ratio; CI, confidence interval.
Risk factors Risk factors and comorbidities are shown in Table II. On univariate analysis, chronic neurological conditions, immunosuppression and chronic heart/metabolic/liver diseases were significantly higher in cases than in controls. However, on multi-variate analysis, chronic neurological, immunosuppressive and cardiac conditions remained significant. Compared with controls, cases were significantly more likely to be
Table I Demographic details of cases and controls Demographics
Cases
Controls
Median age in 45.6 46.8 months (range) (0.9e178.1) (0.7e173.5) Male:female 31:16 93:48 Mean body weight 18.1 (16.7) 17.8 (22.4) in kg (SD) Seasonal flu vaccination 3 (7.5)a 3 (2.5)a rate in 2009, N (%) Ethnicity, N (%) Caucasian 29 (61.7) 78 (55.3) Aboriginal and Torres 3 (6.4) 4 (2.8) Strait Islander Other 15 (31.9) 59 (41.8)
P-value 0.95 1.0 0.24 0.17
exposed to household smokers on both univariate and multivariate analyses.
Clinical presentation The clinical features of both groups are summarized in Table III. Although cough, coryza and fever were the most common presenting symptoms, 38% of children with nosocomial influenza did not have any of these symptoms, compared with 21% of children with community-acquired influenza (P ¼ 0.02). On univariate analysis, cough, subjective fever, lethargy and myalgia/arthralgia were significantly less common in children with nosocomial influenza than in children with community-acquired influenza. However, on multi-variate analysis, only subjective fever and lethargy remained significant. All other clinical symptoms of influenza tended to be less common in children with nosocomial influenza than in children with community-acquired influenza, but this difference was not significant.
Course and outcome 0.44 0.24 0.23
SD, standard deviation. a Forty children in the case group and 120 children in the control group were eligible for influenza vaccine.
The courses and outcomes of the illness have been summarized for both groups in Table IV. The most common complication of pH1N1 influenza was pneumonia, which affected just under 20% of children in each group. A higher proportion of children with nosocomial pH1N1 had bloodculture-proven bacterial co-infection during their course of
G. Khandaker et al. / Journal of Hospital Infection 82 (2012) 94e100 Table III Comparison of clinical features in cases and controls: univariate and multi-variate analyses Clinical features
Cases N (%)
Cough 23 Coryza 21 Fever (documented) 14 Fever (self-reported) 10 Dyspnoea/shortness 9 of breath Vomiting 8 Lethargy 6 Diarrhoea 5 Sore throat 2 Headache 2 Seizure 2 Myalgia/joint pain 0 Absence of ‘cough, 18 coryza and fever triad’ Mean highest recorded 15.3 C-reactive protein Multi-variate analysis Cough Fever (self-reported) Vomiting Lethargy Sore throat Myalgia/joint pain
Controls N (%)
P-value
(48.9) (44.7) (29.8) (21.3) (19.1)
98 69 53 70 36
(69.5) (48.9) (37.6) (49.6) (25.5)
0.01 0.61 0.33 <0.001 0.37
(17.0) (12.8) (10.6) (4.3) (4.3) (4.3) (0) (38.3)
40 55 20 18 13 11 12 28
(28.4) (39) (14.2) (12.8) (9.2) (7.8) (8.5) (20.6)
0.12 <0.001 0.54 0.10 0.28 0.41 0.04 0.02
(SD 65.0) 8.1 (SD 28.7)
Matched OR (95% CI) 1 0.3 0.8 0.3 0.3 0.03
(0.4e2.6) (0.1e0.8) (0.3e2.3) (0.1e0.9) (0.1e1.6) (0.01e8.2)
0.1
P-value 0.99 0.02 0.71 0.03 0.16 0.98
SD, standard deviation; OR, odds ratio; CI, confidence interval.
illness (6.4% vs 1.4%, P ¼ 0.09). Co-infection with other viruses was found in six (4.3%) patients in the community-acquired group (respiratory syncytial virus, adenovirus and parainfluenza virus), and one (2.1%) patient in the nosocomial group (respiratory syncytial virus). A higher proportion of children in the nosocomial group received treatment with oseltamivir (77% vs 43%, P < 0.001) and antibiotics (75% vs 55%, P ¼ 0.02) compared with children in the community-acquired group, although only treatment with oseltamivir remained significant on multi-variate analysis (matched OR 3, 95% CI 1.4e7.5). The use of antibiotics occurred during the course of influenza infection; however, the indication may or may not be related to influenza or its complications. Three-quarters of children in both groups received acetaminophen, and approximately 30% of children in both groups received non-steroidal anti-inflammatory drugs (NSAIDs). In the community-acquired group, three boys aged four, nine and 12 years died during the period of hospitalization. The four-year-old had a brainstem glioblastoma; the nine-year-old had spastic quadriplegia, developmental delay and epilepsy; and the 12-year-old had no known pre-existing medical condition but died one day after hospitalization. All three deaths were attributed to influenza. There were no hospital deaths in the nosocomial group. However, one child died three months after discharge from hospital. He was a two-month-old boy admitted with cholestatic jaundice and failure to thrive, and remained in hospital
97
Table IV Comparison of course of illness and outcomes in cases and controls: univariate and multi-variate analyses Course and outcome Length of hospital stay (days) Length of post-influenza hospital stay (days) ICU admission Length of stay in ICU (days) Ventilated Duration of ventilation (days) Treated with oseltamivir Treated with antibiotics Received NSAIDs Received acetaminophen Developed pneumonia Bacterial co-infectiona Other complicationsd
Cases N (%)
Controls N (%) P-value
13.4 (SD 12.2) 4.5 (SD 8.5) <0.001 8.5 (SD 14.5) 4.5 (SD 8.5)
0.006
7 (14.9) 9.1 (SD 8.6)
0.43 0.95
15 (10.6) 7.0 (SD 8.4)
5 (10.6) 11 (7.8) 3.2 (SD 1.7) 11.2 (SD 16.9)
Multi-variate analysis Treated with oseltamivir Treated with antibiotics
0.37 0.21
36 (76.6)
60 (42.6)
<0.001
35 (74.5)
77 (54.6)
0.02
15 (31.9) 38 (80.9)
39 (27.6) 104 (73.8)
0.58 0.63
8 (17.0) 3 (6.4)b 6 (12.8)
26 (18.4) 2 (1.4)c 15 (10.6)
0.83 0.09 0.67
Matched OR (95% CI)
P-value
3 (1.4e7.5) 1.4 (0.6e3.1)
0.006 0.48
ICU, intensive care unit; NSAIDs, non-steroidal anti-inflammatory drugs; SD, standard deviation; OR, odds ratio; CI, confidence interval. a Numbers are too small to be included in multi-variate model to obtain a meaningful result. b Identified bacteria were Escherichia coli, Acinetobacter spp. and Micrococcus spp. c Identified bacteria were Haemophilus influenzae and Streptococcus pneumoniae. d Pleural effusion, pneumothorax, acute respiratory distress syndrome, convulsion and encephalitis.
for approximately one month while he had pH1N1 infection. He died three months and eight days after discharge from hospital due to causes unrelated to pH1N1. Data on influenza contacts were available for 42 (of 47) children with nosocomial influenza and 105 (of 141) children with community-acquired influenza. Only five of 42 (12%) patients with nosocomial influenza had an identified contact: two children reported contact with other influenza-positive patients, two children reported contact with a symptomatic family member, and one child reported contact with an ill school friend. It was assumed that contact with hospital staff and/or visitors was important for the others. Thirty-six of 105 (34%) children with community-acquired influenza had an identified contact: the majority (83%) reported contact with symptomatic family members, and the rest reported contact with school friends with influenza-like illness.
Discussion This study has described the largest known cohort of children (N ¼ 47) with nosocomial influenza in a single season, and
98
G. Khandaker et al. / Journal of Hospital Infection 82 (2012) 94e100
the first nested caseecontrol study on nosocomial vs community-acquired pH1N1. The only other large cohort of hospital-acquired influenza has been described from Philadelphia Children’s Hospital, involving 46 children aged 21 years hospitalized over four influenza seasons from 2000 onwards.6 This study found that nosocomial influenza differed significantly from community-acquired influenza in terms of disease susceptibility and severity, clinical presentation and aspects of the course of illness. Eighty-one percent of children with nosocomial influenza had an underlying chronic medical condition, compared with 42% of children with communityacquired influenza. Multi-variate analysis showed that chronic neurological, immunosuppressive and cardiac conditions were independent predictors of nosocomial infection, but chronic metabolic, hepatic and pulmonary diseases other than asthma lost significance (Table II). Additionally, children with nosocomial pH1N1 had, on average, a greater number of underlying medical illnesses than children with communityacquired pH1N1 (1.5 vs 0.6, P ¼ 0.03). This is consistent with other studies. Enstone et al. reported that 93% of UK children with nosocomial pH1N1 had serious underlying illnesses, while Fanella et al. from Canada reported that all nosocomial cases had underlying medical conditions.13,14 Both host and environmental factors could be responsible. Generalized immunosuppression, poor functioning of a specific organ/system (e.g. respiratory or cardiac disease) or a combination of both may render these children more susceptible to influenza. Environmentally, as these children require frequent hospitalization for extended periods and need multi-disciplinary care by multiple healthcare workers (HCWs), they may be exposed regularly to patients, visitors or HCWs with influenza. Chronic neurological, immunosuppressive, cardiac and metabolic conditions were the most common pre-existing medical conditions identified in this study, and this is comparable with findings from other studies.6,14 This study also found that children with nosocomial pH1N1 were significantly more likely to have a smoker in the household compared with children with community-acquired disease. Active smoking is known to increase the risk of influenza,17e20 and passive smoking is known to increase the risk of contracting lower respiratory tract infections requiring hospitalization in early life.21,22 However, the mechanism for the increased risk of influenza due to passive smoking or exposure to smokers has not been described in detail to date. Experimental studies suggested that smoke exposure impairs influenza-induced mucosal innate immunity (e.g. through suppression of interleukin-6).18,23 It is also likely that parents who smoke, who tend to cough more than parents who do not smoke, may spread the droplets containing influenza virus through coughing during their visits to their hospitalized children. In this study, children with nosocomial influenza had significantly fewer classical symptoms of pH1N1, particularly subjective fever and lethargy, compared with children with community-acquired influenza.24 Eighty percent (12/15) of UK children with nosocomial pH1N1 presented with fever, but other symptoms were not so common.14 Immunosuppression and concurrent use of antipyretics, analgesics and other medications may modify the classical presentation of influenza in children hospitalized with pre-existing chronic conditions. Babcock et al. commented that patients acquiring influenza while hospitalized were less likely to present with typical
symptoms of influenza, possibly because of concurrent medications with anti-inflammatory and antipyretic effects.25 In the present study, 81% of children with nosocomial influenza received acetaminophen and 32% received NSAIDs during the course of their hospitalization. Interestingly, a similar proportion of children with community-acquired influenza also received these drugs, so the difference cannot be explained purely by the use of anti-inflammatory or antipyretic medications. It may be a special feature of nosocomial influenza to have ‘occult’ or ‘mild’ presentation. Goldwater et al. reported that a respiratory virus (including influenza) or mycoplasma was detected in 46% of children with nosocomial respiratory infection in whom symptoms were initially overlooked, and a respiratory virus or mycoplasma was detected in another 47% of asymptomatic children.5 Influenza is generally a mild illness and only a small proportion of community-acquired disease is sufficiently severe to require hospitalization. As children with nosocomial influenza were inpatients for other reasons, heightened awareness during the pandemic season may have allowed easy detection of milder cases. The UK surveillance study may have underestimated the actual burden of nosocomial influenza. It reported that only 2% of cases of hospitalized children with pH1N1 met the criteria of nosocomial infection.14 In a retrospective analysis of five years of data from a Canadian paediatric hospital, Slinger and Dennis demonstrated that nosocomial influenza accounted for 7.6% of all hospitalized cases of influenza, which is closer to the present finding of 9.3%.4 Therefore, it is possible that the UK surveillance study only identified the more severe pH1N1 cases (there were high rates of death or admission to intensive care), and missed the asymptomatic and less symptomatic cases, resulting in an overestimated case fatality ratio of 20%.14 Such a high case fatality ratio was not seen in other studies.10e13 The present authors believe that if well-designed prospective surveillance is in place, more cases will be found, many of which may be intrinsically mild; if such patients have already been discharged, they would not be ill enough to need to return to hospital. A high index of suspicion is needed to identify influenza in hospitalized children with underlying risk factors to prevent transmission and enable early treatment. More children with nosocomial pH1N1 had blood-cultureproven bacterial co-infection compared with children with community-acquired pH1N1 (6.4% vs 1.4%, P ¼ 0.09). Otherwise, complications in both groups were similar, with just under 20% of children in each group developing pneumonia. This is approximately half of what has been reported from the UK.14 The higher proportion of antibiotic use and insignificantly higher levels of C-reactive protein found in the children with nosocomial influenza may be due to the higher proportion of bacterial co-infection in these children compared with the children with community-acquired influenza. This is supported by a study from Western Australia.26 As expected, the mean duration of hospitalization was longer in the children with nosocomial influenza (13.4 vs 4.5, P < 0.001), as was the length of post-influenza hospital stay (8.5 vs 4.5, P ¼ 0.006). Otherwise, the course of illness was essentially similar in both groups. Understandably, the use of antiviral therapy within two days of onset of illness was higher in children with nosocomial pH1N1 than in children with community-acquired pH1N1 (matched OR 3, 95% CI 1.4e7.5) given their other illnesses. It is not clear if prompt antiviral
G. Khandaker et al. / Journal of Hospital Infection 82 (2012) 94e100 therapy decreased the severity of illnesses in children with nosocomial influenza, but antiviral drug usage in this cohort was comparable with published data.13,14 A limitation of this study is that contact tracing was not performed actively. Only five of 42 (12%) patients with nosocomial pH1N1 whose data were available knew of a contact. Fanella et al. reported that almost all patients with nosocomial pH1N1 acquired the illness from another patient on the ward in their hospital in Canada.13 It is notable that HCWs are also important transmitters of the infection.1,7 Molecular genotyping can help describe the outbreaks of nosocomial pH1N1, although its wider applicability requires further assessment.27 Finally, it is important that all available preventive measures, whether pharmaceutical or non-pharmaceutical, are used to prevent nosocomial influenza in children with atrisk medical conditions. These include annual influenza vaccination of eligible children (children aged six months or more with a chronic medical condition predisposing to severe influenza illness that requires regular medical follow-up or hospitalization), vaccination of HCWs, use of personal protective measures, hand hygiene and tight hospital infection control measures.7 A recent study demonstrated the role of neuraminidase inhibitors in preventing the spread of nosocomial influenza between patients.28 In summary, nearly one-tenth of hospitalized children with pH1N1 in Australia acquired the infection in healthcare settings. Children with underlying medical conditions were at greater risk of developing nosocomial pH1N1. These children may have an ‘occult presentation’ of influenza, and therefore may be considerably under-reported. It is suggested that nosocomial influenza may be far more common than clinicians realize, and should be considered particularly during the influenza season. Conflict of interest statement RB has received financial support from CSL, Sanofi, GlaxoSmithKline, Novartis, Roche and Wyeth to conduct research and attend and present at scientific meetings. Any funding received is directed to an NCIRS research account at The Children’s Hospital at Westmead and is not personally accepted by RB. HM has participated as a board member of global advisory boards for Merck and GlaxoSmithKline, and has participated as an investigator in industry-sponsored clinical vaccine trials. JB has served on advisory and/or data safety monitoring committees for CSL Vaccines and GlaxoSmithKline, for which the Murdoch Children’s Research Institute receives payments into an educational fund. The other authors declare that they have no conflict of interest in relation to this work. Funding sources This project was funded, in part, by the New South Wales Department of Health and the National Health and Medical Research Council (NHMRC) H1N1 Grant No: 633028. Activities of the APSU and NCIRS are supported by the Australian Government Department of Health and Ageing. APSU is supported by the NHMRC (Enabling Grant No: 402784 and Practitioner Fellowship No: 457084, E. Elliott), NHMRC Career Development Fellowship (No. 1016272 H. Marshall), the Discipline of Paediatrics
99
and Child Health and Faculty of Medicine, University of Sydney; the Children’s Hospital at Westmead; and the Royal Australasian College of Physicians.
References 1. Maltezou HC, Drancourt M. Nosocomial influenza in children. J Hosp Infect 2003;55:83e91. 2. Hall CB, Douglas Jr RG. Nosocomial influenza infection as a cause of intercurrent fevers in infants. Pediatrics 1975;55: 673e677. 3. Cunney RJ, Bialachowski A, Thornley D, Smaill FM, Pennie RA. An outbreak of influenza A in a neonatal intensive care unit. Infect Control Hosp Epidemiol 2000;21:449e454. 4. Slinger R, Dennis P. Nosocomial influenza at a Canadian pediatric hospital from 1995 to 1999: opportunities for prevention. Infect Control Hosp Epidemiol 2002;23:627e629. 5. Goldwater PN, Martin AJ, Ryan B, et al. A survey of nosocomial respiratory viral infections in a children’s hospital: occult respiratory infection in patients admitted during an epidemic season. Infect Control Hosp Epidemiol 1991;12:231e238. 6. Leckerman KH, Sherman E, Knorr J, Zaoutis TE, Coffin SE. Risk factors for healthcare-associated, laboratory-confirmed influenza in hospitalized pediatric patients: a caseecontrol study. Infect Control Hosp Epidemiol 2010;31:421e424. 7. Salgado CD, Farr BM, Hall KK, Hayden FG. Influenza in the acute hospital setting. Lancet Infect Dis 2002;2:145e155. 8. Serwint JR, Miller RM. Why diagnose influenza infections in hospitalized pediatric patients? Pediatr Infect Dis J 1993;12: 200e204. 9. Perez-Padilla R, Fernandez R, Garcia-Sancho C, Franco-Marina F, Mondragon E, Volkow P. Demand for care and nosocomial infection rate during the first influenza AH1N1 2009 virus outbreak at a referral hospital in Mexico City. Salud Publica Mex 2011;53: 334e340. 10. Cunha BA, Thekkel V, Krilov L. Nosocomial swine influenza (H1N1) pneumonia: lessons learned from an illustrative case. J Hosp Infect 2010;74:278e281. 11. Redelman-Sidi G, Sepkowitz KA, Huang CK, et al. 2009 H1N1 influenza infection in cancer patients and hematopoietic stem cell transplant recipients. J Infect 2010;60:257e263. 12. Chironna M, Tafuri S, Santoro N, Prato R, Quarto M, Germinario CA. A nosocomial outbreak of 2009 pandemic influenza A(H1N1) in a paediatric oncology ward in Italy, Octobere November 2009. Euro Surveill 2010;15:19454. 13. Fanella ST, Pinto MA, Bridger NA, et al. Pandemic (H1N1) 2009 influenza in hospitalized children in Manitoba: nosocomial transmission and lessons learned from the first wave. Infect Control Hosp Epidemiol 2011;32:435e443. 14. Enstone JE, Myles PR, Openshaw PJ, et al. Nosocomial pandemic (H1N1) 2009, United Kingdom, 2009e2010. Emerg Infect Dis 2011;17:592e598. 15. Elliott EJ, Zurynski YA, Walls T, Whitehead B, Gilmour R, Booy R. Novel inpatient surveillance in tertiary paediatric hospitals in New South Wales illustrates impact of first-wave pandemic influenza A H1N1 (2009) and informs future health service planning. J Paediatr Child Health 2011;23:1440e1754. 16. National Health and Medical Research Council. The Australian immunisation handbook. 9th ed. Canberra: Australian Government Department of Health and Ageing; 2008. p. 184e194. 17. Ward KA, Spokes PJ, McAnulty JM. Caseecontrol study of risk factors for hospitalization caused by pandemic (H1N1) 2009. Emerg Infect Dis 2011;17:1409e1416. 18. Noah TL, Zhou H, Monaco J, Horvath K, Herbst M, Jaspers I. Tobacco smoke exposure and altered nasal responses to live attenuated influenza virus. Environ Health Perspect 2011;119: 78e83.
100
G. Khandaker et al. / Journal of Hospital Infection 82 (2012) 94e100
19. Arcavi L, Benowitz NL. Cigarette smoking and infection. Arch Intern Med 2004;164:2206e2216. 20. Muscatello DJ, Barr M, Thackway SV, Macintyre CR. Epidemiology of influenza-like illness during pandemic (H1N1) 2009, New South Wales. Australia. Emerg Infect Dis 2011;17:1240e1247. 21. Huttunen R, Heikkinen T, Syrjanen J. Smoking and the outcome of infection. J Intern Med 2011;269:258e269. 22. Strachan DP, Cook DG. Health effects of passive smoking. 1. Parental smoking and lower respiratory illness in infancy and early childhood. Thorax 1997;52:905e914. 23. Gualano RC, Hansen MJ, Vlahos R, et al. Cigarette smoke worsens lung inflammation and impairs resolution of influenza infection in mice. Respir Res 2008;9:53. 24. Khandaker G, Zurynski Y, Lester-Smith D, et al. Clinical features, oseltamivir treatment and outcome in infants aged <12 months
25.
26.
27.
28.
with laboratory-confirmed influenza A in 2009. Antivir Ther 2011;16:1005e1010. Babcock HM, Merz LR, Fraser VJ. Is influenza an influenza-like illness? Clinical presentation of influenza in hospitalized patients. Infect Control Hosp Epidemiol 2006;27:266e270. Ingram PR, Inglis T, Moxon D, Speers D. Procalcitonin and C-reactive protein in severe 2009 H1N1 influenza infection. Intensive Care Med 2010;36:528e532. Rodriguez-Sanchez B, Alonso M, Catalan P, et al. Genotyping of a nosocomial outbreak of pandemic influenza A/H1N1 2009. J Clin Virol 2011;1:1. Higa F, Tateyama M, Tomishima M, et al. Role of neuraminidase inhibitor chemoprophylaxis in controlling nosocomial influenza: an observational study. Influenza Other Respir Viruses 2012;6: 299e303.