Uncovering the burden of Influenza in children in Portugal, 2008–2018

Excluding the 2009/10 H1N1 pandemic, the mean annual number of hospitalizations per season was 189, corresponding to 41.3 cases per 100,000 children, which is bellow general estimates for high-income countries [22]. Influenza hospitalization rates for Portugal were approximately half the estimates published for France [9] and Poland [23] – where differences may be partially explained by the included seasons and by cases treated in the emergency services, which were not accounted for in our study [9, 23] – and were also lower than the ones reported for Spain using the same methodology and period of analysis [11]. Future studies should explore the possible causes for these lower rates.

Three-fourths of hospitalized children were aged < 24 months. Unsurprisingly, hospitalization rates decreased with increasing age [24, 25]. Across all age groups, we found that most children hospitalized for influenza were previously healthy, which is consistent with studies from other European countries [9, 25]. The presence of comorbidities increased with age, ranging from 7.8% in children aged 0–5 months to 23.2% in children aged 48–59 months of age.

The study reveals a non-negligible use of mechanical ventilation during hospitalizations coded as due to influenza (2.4% invasive and 3.1% non-invasive ventilation), potentially reflecting the cases requiring care at the ICU. These are not far from those reported by Tillard et al. (2022) for France, where 2.2% of children aged < 24 months hospitalized for influenza were admitted to ICU [9]. Mean in-hospital case fatality risk was also within reported ranges [11, 23]. [14]

The study highlights the complexity of estimating the burden of influenza in children. [28]The total number of hospitalizations coded as due to influenza (2,319) over ten seasons was higher than the estimated influenza-associated excess hospitalizations in all groups of diagnoses – which ranged from 1,760 to 1,879 -, as the models were not able to detect influenza-associated excess hospitalizations in all seasons. Similar conclusions would be reached if we contrasted other published estimates with the hospitalizations coded as due to influenza obtained from our study, for the comparable years [26]. As expected, the excess morbidity and mortality estimates varied across the ten seasons and four groups of diagnoses. More studies are needed to establish which cause-specific data are more adequate when estimating morbidity and mortality due to influenza in children [27] and weekly ILI incidence rates specific for children aged < 5 years old are needed to improve the models’ performance, as different weekly trends in the virus circulations are expected across age groups [28].

Findings from our study raise the question on the extent of the under-detection of influenza amongst the younger Portuguese population. In 2015/16 and 2017/18 we estimated around 650 annual influenza-attributed excess R&C hospitalizations, almost the double from the ones coded as due to influenza, results which appear more aligned with international estimates on the rate of influenza hospitalizations [9, 22, 23]. Furthermore, all models estimated a higher number of influenza-associated excess deaths – ranging from 9 to 95 over the total period – in comparison with the four deaths that had been observed during the hospitalizations coded as due to influenza.

[29]In terms of mortality, we have estimated a mean of 2.1 influenza-associated all-cause deaths per children aged < 5 years old¹, which is lower than the one estimated by Nunes et al. (2.6) between 1980 and 2004 [29]. It is however higher than the one reported more recently by Nielsen et al. in the pooled estimates of mortality attributable to influenza in children aged < 5 years old in 24 European countries, which varied between 0.4 and 1.1 [30], and higher than the 1.5 reported by Pumarola et al. (2023) for Spain, using the same methodology and period of analysis [11]. Comparison of results with those from other studies is challenging as there are few studies reporting results of excess hospitalization and mortality modelling in children – in Portugal and in other European countries – and as they are sensitive to methodological and temporal differences.

Future studies would benefit from better surveillance data in this age group, with age-stratified data and with information on whether hospitalized children had been vaccinated. Generating more comprehensive evidence contributes to a better understanding of the implications that influenza may have to public health and may help to generate awareness and strengthen preventive policies.

[4, 33‐37]Although the mean cost of hospitalization was fourfold in children with comorbidities than in those without them, the total cost to the NHS of influenza hospitalizations was mainly driven by children without comorbidities. These results should be considered when assessing whether the current immunization strategies – targeting only high-risk children in Portugal - are sufficient to address the clinical and economic burden of influenza infection across all children and other indirectly affected age groups. Studies demonstrate that the vaccine effectiveness against influenza-related hospitalizations is high in children [31]. Moreover, children – especially schoolchildren, can also be a vehicle of transmission of the influenza virus across the community [3, 32]. Understanding the impact and cost-effectiveness from the extension of influenza vaccination to all children aged 6 months to 5 years of age put in place in the USA, Canada and by several European countries would be advisable [5, 7, 9, 33].

The results reported in this study should be interpreted considering its limitations. The administrative data from hospitalizations did not include laboratory data nor information on whether the children had been vaccinated for influenza. It is also subject to coding errors or missing information. Hospitalization data includes all NHS hospitalizations, but not those in the private sector. However, the mortality data are representative of the country, as all deaths registered in the nation were used.

The excess morbidity and mortality modeled with ILI cannot strictly be attributed to influenza alone as our time-series models used ILI incidence as an indicator of influenza activity, without additional control variables [18, 34]. [41]Other variables could have also been used, such as temperature and laboratory data on the circulation of other respiratory viruses [35, 36]. Future studies would benefit from considering other respiratory viruses, namely the respiratory syncytial virus which was estimated by Rodrigues et al. (2019) to have an impact on children all-cause mortality in Portugal twice as high as the one from influenza in season 2016/2017 [37]. Furthermore, as previously addressed, the age-stratified weekly ILI incidence rates were not available for the complete study period limiting the predictive capability of the models [28]. Another limitation from our study is that the age groups used in our time series models were not stratified for 0–23 and 24–59 months of age, although the first group represents most of the hospitalizations coded as influenza [38].

Regarding the economic burden of influenza, our study covers only the tip of the iceberg, as only direct costs of hospitalizations in the NHS were considered, thus excluding children treated in the private and in the outpatient setting, both increasing in recent years. Indirect costs of lost productivity from parents or caregivers were not computed, although they may be generated even by milder cases that do not require hospitalization [39].