Pandemic A/H1N1v influenza 2009 in hospitalized children: a multicenter Belgian survey

On March 2009, in Mexico, a novel recombinant influenza strain (A/H1N1v) of swine origin was discovered as an infective agent in humans [1]. This new virus spread rapidly, first to USA and Canada, then all over the world, causing the "new 2009 influenza A/H1N1v pandemic" [2]. Worldwide, the burden of disease was significant and subsequent efforts from health care systems were required to face an overload of patient's consultations as well as to implement vaccination and surveillance programs. Although consequences of the pandemic were less dramatic than initially feared, the World Health Organization (WHO) estimated that the virus was responsible of at least 17700 deaths worldwide and the Centers for Disease Control and Prevention (CDC) reported 59 millions infected people in the USA [3–5]).

During this A/H1N1v flu wave, children and young adults were identified as a particular risk group. They presented a higher attack rate than older adults [6] and a greater mortality rate than previously observed with classical seasonal flu [7, 8]. Several reports on influenza A/H1N1v in pediatric settings have now been published [9–12], but information on clinical presentation and severity of infection in European children remains limited [13–15]. However, these data could be of the great interest to guide future recommendations for vaccination and antiviral therapy during forecoming flu seasons.

Belgium experienced the influenza A/H1N1v epidemic from July 2009 to January 2010. Pandemic vaccine (adjuvanted Pandemrix^®) was only available after the peak occurred in October and was given with priority to risk groups (defined as health care workers, pregnant women and chronically ill patients) [16]. According to our national surveillance system, around 214531 people were infected, 733000 could benefit from vaccination and 19 deaths were attributable to the virus [17].

In this context, we conducted a multicenter study analyzing influenza A/H1N1v pediatric cases hospitalized in four tertiary medical centers of Brussels, Belgium. Our study aimed to offer a comprehensive description of influenza A/H1N1v infection in children, in the light of other recently published data from countries experiencing different health care systems [9–12].

Even though consequences were less dramatic than initially feared, the 2009 influenza A/H1N1v pandemic has caused a significant burden of disease worldwide, especially in the pediatric population [6–8]. Higher attack rate was observed among children, causing important overload in outpatient and emergency departments [11] as well as in PICU [22]. Unfortunately, our study was not designed to assess the epidemiological impact of influenza A/H1N1v infection over the whole pediatric population of Brussels. Moreover, by selecting only laboratory confirmed infections, we underestimated the number of hospitalized cases, especially since diagnosis confirmation by PCR was no more routinely performed at the end of the epidemic. However, we were able to notice an important pediatric burden of disease in Brussels, as illustrated by an increased rate of outpatients visits of 18% during the epidemic period compared to the three previous years and a high rate of PICU occupation by influenza A/H1N1v infected patients (8% during the peak of epidemic). It would have been of interest to compare the rate of hospitalization related to H1N1 with those registered for seasonal flu during the 3 previous years but these data were unfortunately not available.

This study offers a comprehensive description of influenza A/H1N1v pattern of infection among Belgian hospitalized children, in the light of recent publications from other continents. Consistently with these reports [9–12, 23, 24], co-morbidities were highly prevalent among influenza A/H1N1v infected hospitalized children (47%). The co-morbidities were not different from those observed during seasonal flu. As described by others [7, 9, 22], the presence of at least one co-morbidity was significantly more frequent in children of more than two years of age and constituted a risk factor for severity of disease, in terms of PICU admissions and case-fatality rate. Furthermore, influenza A/H1N1v illness course differed according to patients' age groups. Indeed, children less than two years of age (46% of the cohort), and especially those less than three months, presented milder patterns of infection and were often hospitalized only for observation of fever without focus. 86% of PICU admissions and all deaths occurred in children over two years of age (with 80% of deaths among children > five years old). Although this issue is conflicting in the medical literature [7, 10], similar findings have been reported in a large series by investigators from the CDC [25]. This observation differs from what is seen during seasonal flu, where young children and especially infants presented a higher mortality-rate compared to older ones [26, 27]. Nevertheless, during this pandemic wave as well as during previous flu seasons, the highest rate of hospitalization was generally reported among young age groups [8, 9, 11, 24]. This was particularly true in our series, as reflected by a median of age of 31 months, which was even younger than among American and Israeli hospitalized children (median age ranging from four to six years) [9, 10]. If unexplained by the severity of infection, this finding probably illustrates differences in clinical practices and hospitalization policies. In Belgium, the National Healthcare System renders the access to inpatients pediatric facilities easy, so that hospitalization of infants presenting with fever without focus, especially those younger than 3 months of age, is largely recommended and not restricted to the most severe cases as in other countries [28, 29].

Although we focused on hospitalized cases, influenza A/H1N1v episodes were mainly self-limited, consisting of febrile respiratory illness and requiring short duration of hospitalization (median 3 days) with or without oxygen supplementation. Initial clinical features did not differ from seasonal flu [30], except for the higher proportion of children presenting with gastrointestinal manifestations, as described in previous studies [12, 23, 24]. This involvement of the gastrointestinal tract could be subsequent to a high rate of influenza A/H1N1v virus replication [31]. Neurological manifestations were also frequent (10% of children) but in contrast with other reports [9, 22] were not correlated with PICU admission or fatality. Finally, more than one third of the whole cohort and 71% of PICU patients had pneumonia confirmed on chest X-rays. Even though only 3 (1%) children had evidence of bacteremia (all due to S. pneumoniae), it seems very likely that a significant proportion of pneumonia, especially those with lobar infiltrates, were associated with bacterial super-infections. According to some series, bacterial super-infections after influenza A/H1N1v episodes were found in about 4% of hospitalized children [11, 12, 23] but reached 20 to 38% among fatal cases [7, 25, 32]. Considering the low rate of positive blood cultures (BC) in pediatric bacterial community-acquired pneumonia (2.5 to 5%) [33, 34], these published rates as well as our data likely constitute an under-estimation, as bacterial pneumonia diagnosis relied on positive cultures from sterile sites or autopsies and as part of children had received antibiotics prior to microbiological documentation. Finally, among our whole cohort, no necrotizing pneumonia, empyema or sepsis due to S. aureus or group A Streptococcus were reported, even though those pathogens are frequently involved in other series [9, 25].

Surprisingly, a majority of children were treated by antibiotics, even after the diagnosis of influenza A/H1N1v infection was obtained. As mentioned above, confirming bacterial pulmonary super-infection after influenza illness is challenging and diagnostic relies more on clinical presumption and unspecific blood results [35]. However, this could only partly explain the high rate of antibiotics use, as only 54% of those children treated by antibiotics presented pulmonary infiltrates. Rate of antibiotics prescriptions was uniform among all age groups and was also high in other pediatric studies [10–12]. The exact reasons sustaining this practice remain unknown but should be worth to investigate in further prospective studies. Contrastingly, our study showed a particularly low percentage of oseltamivir prescriptions. Indeed, only 24% of the children were treated compared with 45 to 84% in other similar studies [8, 9, 11, 12]. In our four centers, the use of oseltamivir was significantly higher in children above 2 years of age and/or suffering from co-morbidities. This more "watch and wait" practice was in line with the restrictive national recommendations issued for oseltamivir pediatric use during the 2009 pandemic wave, which suggested cautious prescription under one year of age, regarding the absence of safety data among infants [16]. Moreover, in Belgium, prescription of antiviral drugs during seasonal flu is very limited and kept for management of severe diseases or immuno-compromised patients [36]. Obviously, the limitations associated with the retrospective design do not allow us to conclude on treatment efficacy. It is however interesting to note that, although oseltamivir was scarcely used, fatality rate and PICU admissions were comparable to the other above-mentioned reports [9, 10, 12].

In Argentina [8], the case-fatality rate of influenza A/H1N1v infected children was 5%, with a global pediatric mortality rate 10 times greater compared to previous flu seasons. National surveys in the United Kingdom [7] and U.S [23] reported also a higher influenza related mortality rate during the pandemic influenza A/H1N1v than observed with seasonal flu. However, consequences in the Northern hemisphere were less dramatic than anticipated. Studies from these countries reported case-fatality rate among hospitalized children ranging from 0.6 to 3% [9–12], similar to our findings (2%). As previously hypothesized [10], these North/South differences in patients' outcome could partly be explained by an easier access to the health care system in Europe, Israel and North America. In Israel [10], as well as in our series, the median duration of symptoms before hospitalization was only 2 days compared to 4 days in Argentina [8]. On another hand, 2 patterns of influenza A/H1N1v related deaths have been described [7]: those occurring after several days of hospitalization in chronically ill patients (80%), in contrast to those observed after acute evolution of viral infection in previously healthy children (20%). Despite a small number of cases, a similar profile seemed to happen in our series, as the five children who died suffered from chronic neurological disorders and one previously healthy child presented a fulminant viral infection causing cardio-respiratory arrest and requiring nine days of ECMO support to be cured.

Although influenza A/H1N1v infections were globally self-limited, pediatric burden of disease was significant. Children of more than 2 years old and/or suffering from chronic co-morbidities were shown at higher risk of severe infection. Compared to other countries experiencing different health care systems, our Belgian cohort was younger and received less frequently antiviral therapy; disease course and mortality were however similar.