2.1 Study Design

The study was a retrospective cohort study using health insurance claims data. Due to the substantial underreporting and under-testing of RSV [20, 21], in addition to a cohort with confirmed RSV, we included two other cohorts: an RSV-possible cohort (including confirmed RSV cases) and an overall ARI cohort (including confirmed RSV and RSV-possible cases). Control cohorts matched independently for each infection cohort. The three cohorts of respiratory infections were defined based on the literature [22, 23] and adapted and validated through expert opinion. The consulted experts (pulmonologists, a general practitioner, and a laboratory physician) are authors of this manuscript. The codes were discussed in detail in a group discussion, where the specific 10th revision of the International Classification of Diseases (ICD-10) codes were evaluated. Under national guidelines for secondary data analyses, ethical approval and consent to participate were not required [24].

2.2 Database

All study data were obtained from the ‘WIG2 database’ [25], which is an anonymized healthcare claims database containing longitudinal data on 4.5 million individuals from all parts of Germany who are insured within the statutory health insurance (SHI) system.

The WIG2 database contains detailed information on outpatient and inpatient care, drug prescriptions, sick leave, and miscellaneous services (e.g., prescription of a mobility device or occupational therapy), as well as the socio-demographic characteristics of the insured and duration of insurance. Medical diagnoses are coded using the International Statistical Classification of Diseases and Related Health Problems, version 10 – German modification (ICD-10-GM) classification. Medical procedures are coded using a national Operation and Procedure Classification System (OPS). Outpatient visits are billed using a national Medical Fee Schedule (EBM), and drug prescriptions are billed based on the International Anatomical Therapeutic Chemical Classification System (ATC).

2.3 Inclusion and Exclusion Criteria

Inclusion criteria for the infected cohorts were age ≥ 18 years in the year of the index quarter and meeting the ICD-10-GM codes specified per cohort (Online Resource 2 Table S1; see the electronic supplementary material) from 2010 to 2019. The focus of the analysis was on episodes of infections rather than patients. Patients might be included and matched multiple times if a new episode of respiratory infection occurred. To consider an episode as new, a disease-free calendar quarter was required after the previous episode. If patients were, over the years, included multiple times with respiratory infections, we assumed that the episodes were independent events as ARIs are common occurrences within the population.

The confirmed RSV cohort included patients with RSV-specific ICD-10-GM codes. The RSV-possible cohort included all patients from the RSV cohort and patients with respiratory infections and ICD-10-GM codes that suggest they possibly had undiagnosed RSV (Online Resource 2 Table S1). The ARI cohort included all patients with ARI-specific ICD-10-GM codes irrespective of potential underlying pathogens; thus, it also included the confirmed RSV and RSV-possible cohorts. ARI ICD-10-GM codes were based on the Robert Koch-Institute (RKI) definitions for ARI surveillance and previously validated [26]. The only ICD-10-GM code added that was not part of the RKI’s ARI ICD-10-GM codes was the RSV-specific code B97.4. Due to the study’s timeframe (2010–2019), the COVID-19 ICD-10-GM codes were not considered. The ICD-10 codes were selected based on expert discussion. All ICD-10-GM codes, and whether an infection was defined as a lower or an upper respiratory tract disease (LRTD or URTD), are in Online Resource 2 Table S1.

Inclusion criteria for the matched control cohort were age ≥ 18 years in the year of the index quarter and absence of any ICD-10-GM code for RSV, possible RSV, or ARI in the index quarter of the matched patient or one quarter prior.

2.4 Matching and Enrollment

A matching approach was used to determine the additional costs and complications due to the infection, since costs and complications can also occur in patients without infections or with different infections. We adopted a 1:1 propensity score matching combined with an exact matching approach to balance group differences between infected and uninfected individuals. As the age groups (i.e., 18–49, 50–59, and ≥ 60 years) were analyzed separately, a separate matching was carried out for each age group. Patients were matched based on the index quarter of the patient with infection. Details of the matching approach are provided in the section ‘Matching approach’ in Online Resource 2 (see the electronic supplementary material). We included four quarters following the index quarter in our analyses to determine disease burden over a longer period, as RSV infections have been shown to have a long-term effect on costs and disease burden [8, 10].

2.5 Costs and HCRU of RSV Infection

The infection-attributable medical costs were estimated with two approaches: an overall cost calculation and a sector-specific cost calculation. HCRU was assessed for each sector separately (i.e., outpatient, inpatient, medications), whereas sick leave days were only available for the outpatient setting and calculated accordingly.

For the overall cost calculation, patients with infections were compared to matched controls for the index quarter and the four following quarters. Thereby, all costs were considered, including inpatient, outpatient, and medications. Overall costs per patient were calculated by summing costs across all sectors, both for the infected cohorts and the non-infected controls, to then determine excess costs.

For the sector-specific cost calculation, the treatment costs for the specific sectors (outpatient, inpatient, medications) were calculated similarly to a study for influenza [27]. This allowed a more detailed cost assessment. Since outpatient costs are billed per quarter, no direct cost information for the specific treatments was available. Therefore, patients with respiratory infections in the outpatient setting were compared to the matched controls.

The outpatient costs and outpatient HCRU were calculated for the index quarter and the following quarter [27]. Outpatient costs, HCRU, and group differences were calculated for all treatments in the outpatient setting including all specialists.

Medication costs were estimated over two quarters considering the relevant ATC codes by comparing the infection and control cohort. In addition, excess antibiotic and steroid prescriptions were assessed by comparing the infection and control cohort within the quarter of the infection. For steroids, only a new prescription was considered, taking into account the two quarters prior to the index quarter as a washout period. The ATC codes are described in Online Resource 2 Table S2 (see the electronic supplementary material).

For inpatient costs, the primary diagnosis for each hospital stay was selected, enabling the description of costs and HCRU in the hospitals without a control group. Furthermore, hospital length of stay (LOS), intensive care unit (ICU) admission rate, ventilation rate, inpatient mortality, and 30-day all-cause mortality after discharge were assessed.

2.6 Complication Rates

The proportions of patients experiencing complications occurring during the index quarter and the following quarters were computed and compared to the proportions in the matched controls. The complications considered were selected based on medical expertise and included pneumonia, acute respiratory distress syndrome (ARDS), bronchitis, hospitalization due to congestive heart failure (CHF), and exacerbation of chronic obstructive pulmonary disease (COPD) or asthma. While for pneumonia, ARDS, and bronchitis, every occurrence (both inpatient and outpatient) was counted as a complication, for CHF, only the diagnosis in an inpatient setting (primary diagnosis) was counted as a CHF decompensation (and therefore as a complication). For the definition of a COPD exacerbation, we followed Vogelmeier et al. [28]. For an asthma exacerbation, a hospitalization or an emergency visit for asthma was required. The complications investigated are in Online Resource 2 Table S3 (see the electronic supplementary material), including the relevant codes to identify complications.

2.7 Subgroup Analysis

Separate matching and analysis were carried out for the age groups 18–49, 50–59, and ≥ 60, and the respective results are presented separately. In addition, within the age groups 50–59 and ≥ 60, we also performed subgroup analyses by dividing the cohorts into patients with URTD and LRTD. For RSV, only LRTD codes are available, while for the other cohorts, LRTD and URTD were separated based on the wording in the ICD-10-GM code and medical expertise (Online Resource 2 Table S1; see the electronic supplementary material).

Subgroup analyses were also performed for patients with comorbidities of interest (definitions based on [29]), immunocompromised patients (definitions based on [30]), and for patients aged ≥ 60 living in long-term care facilities (LTCF) (identified by EBM codes 03062, 03063, 38100, 38105, 38200). Since an exact matching was performed for these subgroups, analyses could be conducted accordingly. Details of the matching approach are provided in the section ‘Matching approach’ in Online Resource 2. Before the analyses, balancing of covariables was checked and found to be good for all subgroups.

2.8 Statistical Considerations

All interval scale measures were reported using summary statistics such as mean, standard deviation (SD), and median where appropriate. All categorical variables were summarized with absolute and relative frequencies.

Cost data were adjusted to the price level of 2019, and the health-specific consumer price index from the German Federal Statistical Office was used to adjust prices [31]. The cost differences between the infected and control cohorts represented the costs attributable to the respiratory infection and were reported with their calculated 95% confidence interval (CI). For the sector-specific analysis in the outpatient setting, the medication costs, and the difference in antibiotic and steroid use, the same approach was taken by comparing the infected and control cohorts and calculating the difference with 95% CI. For the sector-specific analysis in the inpatient setting, only the primary ICD-10 code for hospitalization was identified, and therefore, all costs and HCRU during the hospital stay were allocated to the primary diagnosis. In this analysis, no comparison with controls was made.

For the assessment of individual complications, and for all cohorts, descriptive analyses were reported in percentages. Group differences were reported with 95% CIs. All patients and their matched controls were considered for the assessment of pneumonia, ARDS, and CHF, while only COPD or asthma patients were considered for COPD and asthma exacerbations. In addition to reporting complication rates, a logistic regression model assessed the increased risk of CHF hospitalization and COPD or asthma exacerbations.

This was done if the analyzed population included ≥ 30 individuals. The results are displayed as odds ratio (OR) and 95% CI. The 95% CIs were considered to determine significance. All analyses were performed using R (version 4.2.0).