During the two 2017-18 and 2018-19 influenza seasons, more than 1000 children under the age of 18 years were hospitalized with influenza in Norway. The majority (75%) of these children were not registered with pre-existing risk conditions for severe influenza. Still, children hospitalized with influenza had more often pre-existing risk conditions compared to the general population within the same age group.
Children under the age of 2 years, and especially those under 6 months of age, were more frequently hospitalized with influenza than older children. Our finding is in line with observations from the United Kingdom (20). Asthma and other lung diseases were the most common risk conditions, both in the general population and among children hospitalized with influenza, and were associated with increased risk for hospitalization. However, more rare risk conditions, such as “compromised immune system” and “epilepsy” were also associated with substantially increased risks of being hospitalized.
We found that 5% of the total Norwegian child population were registered with one or more risk conditions, which is in line with estimates from other studies. A Swedish report, which also used hospital data to calculate the prevalence of risk groups in the population, found that approximately 7.5% of children had one or more risk condition, and a study from the United Kingdom estimated that 3.2% of children aged 0-14 years had a risk condition (21, 22). Higher numbers have been found in a US study, in which it was estimated that 4-12% of children below 18 had one or more risk condition (23), while another study with self-reported data found that 7-14% of all children had a risk condition for severe influenza infection (24).
Previous studies have shown substantial variations in the proportion of children hospitalized with influenza who had pre-existing risk conditions. Estimates range from 12.9% in Austria, 35% in Ireland and 43% in both Bavaria, Germany and Philadelphia, US (13, 15, 25-29). During the influenza A(H1N1) pandemic in 2009, it was estimated that 31% of hospitalized patients globally (among all ages), had one or more risk condition (30). Children have a lower prevalence of chronic diseases than a combined prevalence estimate for all ages, as several established risk conditions, such as chronic heart and lung conditions are much more common among older patients.
One explanation for the relatively high proportion of hospitalized children with no pre-existing risk condition could be that there is a low threshold for hospitalization of children with influenza in Norway. Children, and in particular the youngest ones, are at high risk of complications from lung infections (31). We observed that the youngest children below 6 months had higher hospitalizations rates than those older than 6 months. Dehydration is a possible complication of influenza and children, in particular new-borns, are more difficult to rehydrate and observe clinically than adults. Because of this, the threshold to hospitalize children is lower than for adults. We found that the average length of hospitalization for influenza for all children (4.6 days) was somewhat shorter than previously found for all ages in Norway (5.4 days)(6). For children below 6 months the average length of stay was even shorter, with 4.0 days. This suggests that the hospitalized children does not require more hospital care than adults and that the children are not more severe sick than adults, and supports the assumption that the threshold for admission to hospitals for children may be lower than for adults.
Another reason why we found a high proportion of children with no pre-existing risk condition could be that children with risk conditions receive adequate treatment for their primary condition, which in turn, could lower their risk for severe influenza and admission to hospital.
Lung condition was the most frequently registered risk condition in our study, and we found an increased risk of hospitalization in children with a lung condition. Associations, although not strong, between asthma and severe outcome of influenza infection have also been found by others (12, 30), but not all. McLean et al. did not find that asthma increased the risk of severe illness in children (32). It is difficult to obtain accurate prevalence estimates of asthma in children, as diagnostic criteria and data sources vary between countries. A Norwegian study based on data from the national prescription database found that 1 in 20 children aged 0-19 years (5%) had used anti-asthmatic drugs for more than three months (33), which is higher than our estimate based on primary care consultation data. However, as influenza can cause worsening of asthma symptoms, some patients may be registered with asthma diagnosis and not an influenza-diagnosis when hospitalized with influenza (34). If so, we have missed some influenza-hospitalizations in this group of children, which may have attenuated our associations between asthma and influenza. This could also be the case for other conditions that may be exacerbated by influenza.
We found that 0.5% of the children in the total population were registered with epilepsy, in agreement with findings from Finland and UK (35, 36). Our results indicate that epilepsy was associated with a substantially higher risk for hospitalization with influenza. Epilepsy has previously been identified as a risk factor for severe influenza infections and death in children (13, 37), supporting our finding of epilepsy as a risk condition for severe influenza. Influenza vaccination is recommended for children with epilepsy due to the well documented increased risk of seizures caused by viral infections, causing fever (38, 39). Our results clearly show an increased risk of influenza hospitalization for children with heart conditions or who were immunocompromised. This finding supports the current recommendation that children with such conditions should be vaccinated. Both immunosuppression and neurological disease have previously been linked with an increased risk of death among patients with influenza infection (40), and Coffin et al. found that neurological disease and cardiac disease caused prolonged hospital stay in those younger than 21 years of age (41).
Due to the low numbers of children identified with diabetes mellitus in our study, we could not evaluate the risk in this group. A more extensive study is needed to be able to obtain risk estimates for this patient group.
Influenza vaccines are not approved for patients younger than 6 months (42). The finding that children under 6 months are overrepresented among hospitalized children provides a strong argument for vaccinating pregnant women. Vaccination of pregnant women will not only protect the mother from influenza complications more often seen in pregnant and post-partum women (12, 30) but also protect the infant the first months after birth by transfer of maternal antibodies to the foetus.
When using registry data to identify risk conditions, there are several limitations. The sensitivity for capturing risk conditions depends on whether all diagnoses on existing conditions are registered during health care visits, and that patients regularly see a doctor for their condition. A Norwegian study in adults found that only 60% of adult patients with diabetes had seen a specialist the last year (43). However, children in Norway with chronic conditions such as diabetes or epilepsy, are very likely to have annual visits and check-ups in specialist care and are thereby expected to be correctly defined in this registry-based study. Our prevalence estimates of the selected risk conditions are in line with other studies, supporting that we have identified most children with known risk conditions in Norway.
The susceptibility of influenza infection in children varies with the type of influenza virus. Surveillance reports show that the 2017-18 season in Norway was dominated by the influenza B Yamagata virus, and the 2018-19 season was dominated by the influenza A(H1N1) virus (44, 45). Influenza B viruses have been seen to cause a higher number of severe infections and fatalities in children compared to influenza A outbreaks (37, 46). This could be reflected in our findings of a slightly higher number of hospitalizations in the 2017-18 season. We did not have information on laboratory confirmation on influenza as the hospital registry does not provide information on laboratory tests. However, some of the ICD-10 codes for influenza do require laboratory confirmation, and tests are increasingly used in Norway when patients are hospitalized with influenza (47). We therefore believe that the risk of misclassifying influenza in hospitalized patients is relatively low. Also, we excluded out-of-season diagnoses to reduce further the probability of including patients with other causes of infection wrongly classified as influenza. Moore et al. found a sensitivity of using ICD-10 codes to identify influenza hospitalizations of 86 % and a very high specificity of 98% (48). It is unlikely that all children hospitalized with influenza have been tested for influenza, which can cause underreporting of the disease. Also, the children could be registered with diagnoses due to complications of their primary condition and not including influenza as additional diagnose. Thus, the number of influenza-attributable hospitalizations in our study could be underestimated.
Another limitation of this study is the lack of information on the vaccination coverage as influenza vaccination will influence the hospitalization rate. However, children in Norway are not routinely recommended seasonal influenza vaccination, and in general, seasonal influenza vaccination rates are low in children. Children with risk conditions are recommended annual influenza vaccinations. Although the vaccination rate is considered to be insufficient (45), vaccination of high-risk children has probably prevented some hospitalizations. For this reason, children in risk groups may have an even higher underlying risk of severe influenza and hospitalization.