Burden of disease
Our study, during the years 2010-2019, analyzes 156,097 reported patients hospitalized in Germany because of a laboratory confirmed Influenza primary diagnosis. Overall, the majority of patients were >59 years (46.9%) and had a J10 ("Influenza due to identified seasonal influenza virus") primary diagnosis (84.6%). Among patients >59 years, 27.7% of those hospitalized had diabetes as a pre-existing condition and 33.1% suffered from pneumonia caused by Iinfluenza as a complication. The youngest (<18 years) and oldest (>59 years) age groups carried the highest burden of Influenza measured by the incidence rates. Out of all the risk factors we selected to evaluate, diabetes was the most common in all years except in 2011, where immune disorder was more common. This can be explained by the fact that patients >59 years (the largest represented group in all years except 2011) have a greater instance of diabetes, but in 2011, the <18 years age group accounted for the majority of the hospitalized patients (54.6%). A study from Von der Beck et al. (2017), also based on the Database of the German Federal Statistical Office, analyzed data of Influenza inpatients from the year 2005 to 2012 . Von der Beck found that younger patients were more frequently hospitalized during the 2009/2010 A(H1N1) influenza pandemic. Our data from 2010 and the post-pandemic years shows the remnants of that pandemic in 2010-2013, where patients <18 years comprised of more than 45% of all Influenza related hospitalizations in Germany. Compared to our study, Von der Beck et al. concluded that very young (0-4 years) and very old (>60 years) patients are less frequently hospitalized during the pandemic, whereas during non-pandemic seasons these age groups bear the highest burden of disease. Our results show that, since 2017, patients >59 years are the majority again and make up more than 50% of all Influenza-associated hospitalizations. The greatest burden of disease regarding frequency of hospitalizations has in the last couple of years, thus, shifted from children (5-14 years)  to older adults (>59 years).
As already observed by Von der Beck, interestingly, a substantial burden also occurred in adults 18-59 years during the pandemic. During the post-pandemic years, they also showed considerable proportions of IAH, with over 30% during the years 2011 and 2014, and usually representing between 20 and 30% of the annual IAH. In the present analysis, adults of this age group have the largest proportion of serious complications such as ARDS (2.9%, N= 1004) and Sepsis (3.3%, N= 1133). Adults (18-59 years) also have the highest proportion of patients that had to receive intensive care treatment (8.4%, N= 2918) and invasive ECMO therapy (1.0%, N= 361). This is particularly relevant since adults in this age group fall outside of the recommendations of seasonal Influenza vaccinations in Germany.
The current analysis describes and quantifies the direct healthcare costs of laboratory confirmed Influenza hospitalizations in Germany from January 2010 to December 2019. The overall mean (SD) annual per patient direct cost of hospitalization was 3521€ (±8886€) while the overall median (IQR) annual per patient cost was 1805€ (1502€; 2694€). In the span of ten years, 156,097 hospitalized patients with a confirmed Influenza infection (ICD-10-GM J09/J10 primary diagnosis) amounted an estimated 549,213,936 EUR in just direct costs.
Patients incurred higher costs when intensive care treatment and/or a complication, especially ARDS, Sepsis, and Pneumonia caused by a virus other than Influenza, was reported. These results correspond with those found by a Karve et al. study that found that healthcare costs among Influenza patients with complications double those of Influenza patients without any, although in our results in some cases with complications the cost was almost 10 times higher than the average median and mean costs for all Influenza hospitalized patients .
Our study includes patients hospitalized due to a laboratory confirmed Influenza diagnosis, that is with the ICD-10-GM code J09 or J10 reported as primary diagnosis. According to Ehlken et al. 2015, over 92% of Influenza attributable episodes are classified as “Influenza, virus not identified” J11 . In our study, we explicitly excluded patients solely with a primary diagnosis of J11 in order to avoid including patients hospitalized with Influenza-like symptoms that could be caused by a number of other viruses, some of which might have different risk factors and complications associated with them.
The 2009/2010 Influenza A(H1N1) pandemic claimed the lives of 151,700-575,400 people worldwide during that influenza season . It is estimated that roughly 80% of all A(H1N1)pdm09-related deaths occurred in people younger than 65 years, which was a shift from the typical influenza mortality rates, which disproportionately affect people older than 65 years. Influenza A(H1N1)pdm09 is only associated with the ICD-10-GM J09 code. Our individual year analysis (Additional Table 2) reveals that in 2010-2011, the pandemic strain was still the most prevalently coded in hospitalized Influenza patients, but in 2012 there is a sudden shift and the proportion of patients with J09 gradually begins to decrease (with reasonable fluctuations) until dropping below 10% in 2018. This decrease in the use of the Influenza code J09 is likely to reflect a change in coding procedures, as the previous ‘pandemic’ Influenza virus A(H1N1)pdm09 is now considered a ‘seasonal’ virus and may therefore be coded increasingly as “J10”. It is noticeable that in 2010, compared to the ten-year average, the total number of persons hospitalized was low (as the main wave of the pandemic occurred in 2009 and, hence was not covered by the present analysis), while mean costs per person were high (Table 4). This may be due to higher prevalence of J09 (Additional Table 2), higher costs associated with J09 (Additional Table 3), and higher prevalence of ECMO treatment (data not shown) in that study year.
Comparison with other studies is difficult due to differences in healthcare system, unit costs of applied resources, case definition, and study population. Our study includes costs that are reported by the hospitals and not estimated. In a study conducted by Haas et al. on 65,826 Influenza patients in a health claims data analysis from the season 2012/2013, it was estimated that in Germany mean Influenza hospitalization costs were roughly 5832€ which is much higher than in our study (3806€) . An explanation for this may be that the Haas et al. study defines influenza hospitalization through the ICD-10-GM codes J09, J10 and J11 either as primary or as any secondary diagnosis. It is possible that relevant chronic conditions or complications coded as primary diagnosis with an influenza code as secondary diagnosis were associated with higher costs than those with a J09/J10 primary diagnosis with the chronic condition or complication stated as secondary diagnosis. In contrast, the estimated average inpatient Influenza case cost between 2012 and 2014 in a further recent analysis of German health claims data by Scholz et al. (for cost calculations reference year 2014) based on Influenza as primary diagnosis was 2033€ (SD±2952€), lower than the costs we calculated for those years . However, their data set contained only 458 hospitalized patients and the case definition included J11 as primary diagnosis ("Influenza, virus not identified") which may be associated with less costs.
Similar to a US study by Young-Xu et al. we found the highest per patient costs in the age groups in the 40-79 age range . Generally, patients older than 90 years old did not incur particularly high costs in our cohort. This is probably because of the high risk and high-cost invasive procedures such as ECMO are less frequently used in very elderly patients. In fact, our study reveals that Influenza patients who had an ECMO procedure while hospitalized incurred costs that were more than 10 times the overall median and mean per patient cost, even though they made up less than 0.5% of all patients.
In our study, patients >59 years have the longest average length of hospital stay (7 days). According to the CDC, older adults tend to have longer stay at the hospital due to complications associated with underlying diseases . We also see in our study that patients in the >59 years age group have higher rates of complications. Sepsis was reported in 3.2% (N= 2,345) of all cases >59 years, whereby patients with a Sepsis complications have a mean (SD) per patient cost of 29,215€ (±36,080€). Sepsis (N= 1,133, 3.3%) and ARDS (N= 1,004, 2.9%) were also reported in 18-59 year olds more frequently than in the other two age groups, the latter having the highest per patient mean (SD) cost out of all the selected complications (38,372€, ±35,996€). We also see that the 18-59 years age group have the largest proportion of patients receiving treatment at the ICU (8.4%) and having ECMO procedures (1.0%) performed. ECMO is a bypass circuit that essentially serves as an artificial membrane lung with patients with severe respiratory complications. Complications on ECMO are not uncommon and the procedure is generally considered a high-risk operation that is reserved for patients with very severe respiratory complications, such as ARDS. Higher rates of ECMO treatment among the 18-59 years age group could be an indication that they are subject to more severe complications or that physicians are more hesitant to use high-risk procedures on children (<18 years) and older adults (>59 years). Nevertheless, ECMO use is still a rare treatment option for hospitalized Influenza patients.
Our results show that substantial burden and costs occur in adults >59 years: they make up 46.9% of all hospitalized Influenza patients and have the overall highest median and mean per patient direct cost. In Germany, the Robert Koch Institute recommends that all adults 60 years of age or older should receive yearly Influenza vaccination [24, 25]. Improving Influenza vaccination uptake may have the potential to reduce this burden . As of 2017, less than 40% of the population older than 59 years of age were vaccinated against the seasonal flu, and that percentage was a slight decrease from the previous year. This is well below the European Union’s target of 75% vaccination coverage . It is now more apparent than ever that in Germany, Influenza vaccine recommendations for adults should be revised in the future and public health efforts in the country need to increase, especially among the elderly population and those with a chronic underlying disease.
Our analysis closely resembles a cost of illness study, where the goal is to quantify the cost, or economic burden, of a specific condition. Cost of illness studies have ultimately a descriptive role and are not a true economic evaluation because there is no analysis of the competing course of action . Therefore, although we are able to extrapolate mean per patient direct costs, ultimately, we are not able to appraise whether these costs are significantly high or low enough to create a negative effect on the economy. However, since we cover rare events with serious economic consequences, cost data (and the annual fluctuation in costs) may contribute information for public health decision making.
It should be noted that our study was focused on direct claim costs for patients. A 2007 study in the US reported that indirect costs (such as productivity losses) were about 10 times higher than direct medical costs . Hence, the total healthcare costs of IAH are most likely far higher.
There are some important limitations to the interpretation of our results. Our study only includes inpatient data, so symptomatic cases who do not seek medical attention or seek outpatient care are not considered for analysis. Also, no out-of-pocket payments or co-payments that are not reimbursed by the German sickness funds are part of our analysis. A study by Molinari et al. in 2007 showed that the actual costs of seasonal Influenza is grossly underestimated since the disease is so underreported . This is not ideal since, as mentioned previously, various studies have reported that direct inpatient medical costs are only a small portion of the total economic burden of disease compared to indirect costs such as work absenteeism. In addition, we have no information on the vaccination status of the patients, so we are not able to see whether the vaccine had any protective effect against the more severe complications, and thereby the cost. Although our study has great external validity since the whole German population is included, there is low internal validity due to potential coding issues. However, to characterize Influenza patients we focused in the present study on the Influenza ICD-10-GM codes used for laboratory-confirmed Influenza, leaving out the code J11 ("Influenza, not laboratory confirmed"). We had previously conducted a sensitivity analysis that showed that J11 as main diagnosis represented only a small proportion of cases (17% of patients with a main diagnosis J09/J10/J11) and had only low impact on the costs.
Furthermore, we focused on only those patients with laboratory-confirmed Influenza codes used as primary diagnosis. Hence, it is likely that the ‘true’ number of hospitalized patients with Influenza is far higher, as e.g. those patients were missed with a primary diagnosis of a likely Influenza-associated complication, e.g. secondary bacterial pneumonia, or myocarditis. Thus, our cost analysis restricted to well-defined IAHs represents only a minimum estimate of the true costs.