Anthropometric data and distribution of infections
Of 921 cases from the internal medicine, neurology and paediatric units, 906 patients (413 women (45.6%), 493 men (54.4%)) remained eligible for analysis, since they were not admitted a second time with the same virus within four weeks. The median age (quartiles) of the study population was 68.0 (0.4; 81.3) years, the BMI 25.7 (22.9; 29.3) kg/m2. The distribution of sex and age according to infection groups is given in Table 1. The majority of patients were infected with SARS-CoV-2, while the second most common virus was RSV, followed by Influenza A. Due to the fact, that the group of patients with Influenza B comprised only 2 patients and the groups with combined infections were very small, all subsequent statistical comparisons were limited to the groups of Influenza A, SARS-CoV-2 and RSV.
Table 1
Baseline characteristics stratified for the type of infection. Numbers (percentages) and median values (quartiles) are given. * Percentages refer to infection group (rows). BMI = Body Mass Index. ** Percentages refer to total group of 906 patients. n.d. = not determined.
Viral infection | n = 906 ** | n (< 18 years)* | Sex (m/f)* | Age (years) | BMI (kg/m2) |
Influenza A | 198 (21.9%) | 51 (25.8%) | 106/92 (53.5%/46.5%) | 64 (15; 78) | 26.0 (23.2; 29.8) |
Influenza B | 2 (0.2%) | 1 (50%) | 1/1 (50%/50%) | 22 (n.d.) | n.d. |
SARS-CoV-2 | 437 (48.2%) | 46 (10.5%) | 243/194 (55.6%/44.4%) | 77 (61; 84) | 25.5 (22.8; 29.1) |
RSV | 256 (28.3%) | 157 (61.3%) | 136/120 (53.1%/46.9%) | 2 (0.4; 70) | 26.0 (22.4; 29.3) |
Influenza A + SARS | 4 (0.4%) | 0 (0%) | 3/1 (75%/25%) | 83 (75; 86) | 25.0 (n.d.) |
Influenza A + RSV | 6 (0.7%) | 3 (50%) | 2/4 (33.3%/66.7%) | 26 (4; 59) | 24.1 (n.d.) |
Influenza B + RSV | 1 (0.1%) | 1 (100%) | 1/0 (100%/0%) | n.d. | n.d. |
SARS + RSV | 2 (0.2%) | 2 (100%) | 1/1 (50%/50%) | 0.15 (n.d.) | n.d. |
Regarding BMI and sex, there were no significant differences between the three groups. For age, however, all three groups were different from each other (p < 0.001 each). Age distribution is illustrated in Fig. 1, showing two distinctive peaks. When restricting the analysis to patients with age ≥ 18 years, age again differed between the three groups (p < 0.001), with median values (quartiles) of 72 (62; 80) for Influenza A, 79 (68; 84) for SARS-CoV-2, and 76 (63; 85) years for RSV. In this case, age was only different between Influenza A and SARS-CoV-2 (p < 0.001), whereas the RSV group did not significantly differ from the other two groups. In patients of age < 18 years, age also differed between groups (p < 0.001), with median values (quartiles) of 3.8 (1.9; 8.4) for Influenza A, 0.51 (0.16; 1.87) for SARS-CoV-2, and 0.49 (0.17; 1.92) years for RSV. The SARS-CoV-2 and the RSV group were not different from each other, but age of both significantly differed from age of Influenza A patients (p < 0.001 each).
Comorbidities
In patients aged ≥ 18 years, heart failure, rheumatic disease, COPD, and a state of immunosuppression were most often present with RSV. In SARS-CoV-2 patients, asthma and COPD showed the least prevalence, peripheral arterial disease (PAD) the highest. Chronic kidney disease (CKD) was least prevalent in Influenza patients. If expressed as sum of the comorbidities given in Table 2, this sum was significantly greater in RSV patients compared to the other two groups (p ≤ 0.003 each).
Table 2
Distribution of comorbidities of patients aged ≥ 18 years. Numbers (percentages) are given, for the sum of all comorbidities median values (quartiles). Statistical comparisons were performed with the chi-square statistics or the Kruskal-Wallis test. COPD = Chronic obstructive pulmonary disease. *except malignant diseases of the lung. **within the last 5 years. *** count of all comorbidities listed in the Table.
Prevalence of comorbidities of patients aged ≥ 18 years |
| Influenza A | SARS-CoV-2 | RSV | p value |
n | 147 | 391 | 99 | - |
Hypertension | 81 (55.1%) | 254 (65.0%) | 64 (64.6%) | 0.098 |
Peripheral arterial disease (PAD) | 4 (2.7%) | 38 (9.7%) | 5 (5.1%) | 0.014 |
Heart failure | 25 (17.0%) | 84 (21.5%) | 34 (34.3%) | 0.005 |
Coronary artery disease (CAD) | 42 (28.6%) | 99 (25.3) | 36 (36.4%) | 0.088 |
Diabetes mellitus type 2 | 30 (20.4%) | 112 (28.6%) | 31 (31.3%) | 0.096 |
COPD | 30 (20.4%) | 43 (11.0%) | 30 (30.3%) | < 0.001 |
Asthma | 25 (17.0%) | 20 (5.1%) | 18 (18.2%) | < 0.001 |
Other lung disease* | 7 (4.8%) | 19 (4.9%) | 5 (5.1%) | 0.995 |
Chronic kidney disease (CKD) | 17 (11.6%) | 86 (22.0%) | 23 (23.2%) | 0.017 |
Active malignant disease** | 12 (8.2%) | 56 (14.3%) | 11 (11.1%) | 0.142 |
Rheumatic disease | 6 (4.1%) | 18 (4.6%) | 11 (11.1%) | 0.028 |
Depression | 9 (6.1%) | 31 (7.9%) | 10 (10.1%) | 0.521 |
Dementia | 18 (12.2%) | 49 (12.5%) | 11 (11.1%) | 0.928 |
State of immunosuppression | 19 (12.9%) | 50 (12.8%) | 22 (22.2%) | 0.049 |
Number of comorbidities*** | 2 (1; 3) | 2 (1; 4) | 3 (2; 4) | < 0.001 |
To clarify, to which extent the differences in prevalence could be attributed to the differences in age, we performed logistic regression analyses with age and the three types of infections as predictors, and the comorbidities as outcomes. Age was significantly (p < 0.05 each) associated with hypertension, PAD, heart failure, coronary artery disease (CAD), diabetes mellitus type 2, asthma, CKD, and dementia but all comorbidities that showed a significant unadjusted difference between infections (Table 2) remained significantly (p < 0.05 each) linked to the different infections, suggesting virus-specific, age-independent risk profiles. In patients of age < 18 years, comorbidities were not analyzed due to lack of data.
Primary outcomes
The distribution of ICU treatment and in-hospital mortality for all patients and all infections is shown in the Supplemental Table S1. Neither the frequency of ICU admission (p = 0.974) nor that of in-hospital mortality (p = 0.109) differed significantly between the three groups Influenza A, SARS-CoV-2 or RSV. To account for the differences in clinical characteristics, the age groups < 18 years and ≥ 18 years were then analyzed separately (Table 3). In adults, the three major infection groups again showed no significant differences regarding ICU admission (p = 0.649) or in-hospital mortality (p = 0.496). Of the younger patients, all survived, and it appeared that ICU admission was more frequent in the RSV group, but due to low case numbers the chi-square statistics was of limited value.
Table 3
Outcome data of the three major infection groups stratified according to age. Numbers (percentages) are given. For the results of statistical comparisons, see text. Mortality refers to in-hospital mortality. ICU = Intensive Care Unit.
Age group | < 18 years | ≥ 18 years |
Infection | n | ICU admission | Mortality | n | ICU admission | Mortality |
Sample size | 254 | 17 (6.7%) | 0 (0%) | 637 | 103 (16.2%) | 56 (8.8%) |
Influenza A | 51 | 2 (3.9%) | 0 (0%) | 147 | 24 (16.3%) | 10 (6.8%) |
SARS-CoV-2 | 46 | 0 (0%) | 0 (0%) | 391 | 60 (15.3%) | 35 (9.0%) |
RSV | 157 | 15 (9.6%) | 0 (0%) | 99 | 19 (19.2%) | 11 (11.1%) |
Treatment characteristics
Table 4 provides data on the treatment characteristics of patients aged ≥ 18 years. The length of the hospital stay differed between the three groups (p = 0.021), with a significant (p = 0.033) difference between SARS-CoV-2 and RSV. There were no significant differences regarding the other durations. The frequencies of NIV, low-flow oxygen supply during the hospital stay and oxygen supply upon admission differed significantly between the three groups (p < 0.001 each), whereby the highest percentages were observed in the RSV group.
Table 4
Treatment characteristics of patients of age ≥ 18 years for the three major infection groups. Numbers (percentages) are given. Statistical comparisons were performed with the chi-square statistics and the Kruskal-Wallis test. Durations refer to the subgroups of patients in whom the respective treatment was applied.
Treatment characteristics of patients aged ≥ 18 years |
| Influenza A | SARS-CoV-2 | RSV | p value |
n | 147 | 391 | 99 | - |
Length of hospital stay, days | 7 (4; 10) | 7 (4; 12) | 6 (3; 9) | 0.021 |
Intensive care unit | | | | |
Frequency | 24 (16.3%) | 60 (15.3%) | 19 (19.2%) | 0.649 |
Length of stay, days | 2.7 (1.0; 6.7) | 2.8 (0.9; 7.1) | 2.1 (1.2; 5.6) | 0.981 |
Mechanical invasive ventilation | | | | |
Frequency | 8 (5.4%) | 17 (4.3%) | 3 (3.0%) | 0.662 |
Duration, hours | 33.6 (11.5; 138.0) | 157.1 (20.8; 298.1) | 57.7 (n.d.) | 0.439 |
Non-invasive ventilation | | | | |
Frequency | 10 (6.8%) | 11 (2.8%) | 12 (12.1%) | 0.001 |
Duration, hours | 7.9 (2.8; 60.5) | 13.3 (7.5; 47.2) | 28.1 (9.6; 56.5) | 0.460 |
Oxygen supplementation | | | | |
High-flow during stay | 13 (8.8%) | 25 (6.4%) | 9 (9.1%) | 0.486 |
Low-flow during stay | 99 (67.3%) | 227 (58.1%) | 82 (82.8%) | < 0.001 |
Upon admission | 46 (31.3%) | 102 (26.1%) | 46 (46.5%) | < 0.001 |
Table 5 shows analogous data for patients aged < 18 years. The length of the hospital stay differed between the three groups (p < 0.001), with significant differences between RSV and SARS-CoV-2 as well as Influenza A (p ≤ 0.003 each). There were no significant differences regarding the other durations. The frequencies of high-flow and low-flow oxygen supply during the hospital stay also significantly differed between groups (p ≤ 0.004 each), whereby the highest percentages occurred in the RSV group.
Table 5
Treatment characteristics of patients of age < 18 years for the three major infection groups. Numbers (percentages) and median values and quartiles are given. n.d. = not determined. Statistical comparisons were performed with the chi-square statistics but partially have to be considered only as hints due to the low case numbers. Durations refer to the subgroups of patients in whom the respective treatment was applied.
Treatment characteristics of patients aged < 18 years |
| Influenza A | SARS-CoV-2 | RSV | p value |
n | 51 | 46 | 157 | - |
Length of hospital stay, days | 2 (1; 5) | 2 (1; 3) | 4 (2; 6) | < 0.001 |
Intensive care unit | | | | |
Frequency | 2 (3.9%) | 0 (0%) | 15 (9.6%) | 0.05 |
Length of stay, days | 4.9 (n.d.) | n.d. | 3.7 (2.4; 7.8) | 0.881 |
Mechanical invasive ventilation | | | | |
Frequency | 0 (0%) | 0 (0%) | 3 (1.9%) | 0.391 |
Duration, hours | n.d. | n.d. | 108.8 (n.d.) | - |
Non-invasive ventilation | | | | |
Frequency | 0 (0%) | 0 (0%) | 3 (1.9%) | 0.391 |
Duration, hours | n.d. | n.d. | 66.9 (n.d.) | - |
Oxygen supplementation | | | | |
High-flow during stay | 3 (5.9%) | 1 (2.2%) | 29 (18.5%) | 0.004 |
Low-flow during stay | 15 (29.4%) | 0 (0%) | 114 (72.6%) | < 0.001 |
Upon admission | 2 (3.9%) | 0 (0%) | 11 (7.0%) | 0.151 |
As the group of very young patients seemed of particular interest, we analyzed data of children aged < 3 years separately (Supplemental Table S2). The majority had RSV infection; these patients again showed the longest duration of their hospital stay and the highest percentage of oxygen therapy, either low-flow or high-flow.
Prevalence of symptoms
Symptoms were analyzed only for patients aged ≥ 18 years. The prevalence of cough, dyspnoea and fever showed significant differences between the three major types of infection (p < 0.001 each), with low values for cough and dyspnoea in SARS-CoV-2, high values for cough and dyspnoea in RSV, and a high value of fever in Influenza A (Supplemental Table S3). To account for a possible dependence on age, again logistic regression analyses were performed including age and the type of infection as predictors, and each symptom as outcome. The prevalence of nausea decreased with increasing age, while that of fatigue increased but the unadjusted differences between virus type (Supplemental Table S3) remained significant, suggesting virus-specific, age-independent patterns of symptoms.
Vital parameters, arterial blood gas and laboratory parameters upon admission
For patients ≥ 18 years, vital parameters upon admission are given in the Supplemental Table S4. All of them, except systolic blood pressure, differed significantly (p < 0.05 each) between infection groups. According to post hoc-comparisons, SARS-CoV-2 and RSV differed regarding respiratory rate, heart rate, oxygen saturation and diastolic blood pressure (p < 0.05 each). For heart rate, temperature and oxygen saturation, there were also significant differences between Influenza A and SARS-CoV-2 but we never observed significant differences between the Influenza A and RSV group. pO2 did not significantly differ between groups, but pCO2 and pH did. Regarding pH, the SARS-CoV-2 group showed higher values than the other two groups; regarding pCO2, values were highest in the RSV group (p < 0.05 each). Among laboratory parameters, only eGFR, CRP and D-dimers were significantly different between groups (p < 0.05 each) and are shown in the table. Specifically, eGFR and CRP differed between Influenza A and SARS-CoV-2 groups, both with higher values for Influenza A. D-dimers differed between SARS-CoV-2 and RSV, with higher values in SARS-CoV-2.
Risk factors for ICU admission and in-hospital death
The descriptive results given above showed similarities between the three infections, but also pointed towards differences. As some of the risk factors, such as age or comorbidities, were linked to each other, we performed multiple logistic regression analyses with the aim to identify the statistically independent predictors of ICU admission or in-hospital death. The initial choice of variables was guided by their potential relevance for the outcomes; in the final set we eliminated all predictors with p-values of 0.10 or higher. However, the three infection categories (Influenza A, SARS-CoV-2, RSV) were always kept as predictors irrespective of statistical significance, with Influenza A as reference, in order to compare their impact with that of other predictors. The approach followed by us is outlined in the Supplement in detail.
Regarding ICU admission, the results and odds ratios of associations are illustrated in Fig. 2, indicating that oxygen supply upon admission was the most consistent, strongest (p < 0.001) predictor of subsequent ICU admission for all three viruses, with an odds ratio (95% CI) of 3.88 (2.35; 6.42). Age, sex, active malignant diseases, heart rate, body temperature and oxygen saturation upon admission were additional predictors, while the type of infection was not significantly associated with ICU admission, in line with the unadjusted comparisons.
Regarding in-hospital mortality, odds ratios are shown in Fig. 3. eGFR and initial oxygen supply were robust and significant predictors (p < 0.05 each), whereas initial oxygen saturation and systolic blood pressure showed only a tendency. The type of infection was not significant, in accordance with the unadjusted comparisons.
Sensitivity analysis
For this purpose, we identified patients, in whom the infection was considered to have been the likely cause of their hospital stay. In patients < 18 years of age, only 24 of 254 were excluded as having incidental positive tests for any of the three viruses. The numbers of ICU admission were unchanged. Regarding in-hospital mortality, there were also no changes as this was already zero.
In patients of age ≥ 18 years, 202 of 637 were excluded (Supplemental Table S5). More SARS-CoV-2 patients were excluded compared to the other two viruses. The relative frequencies of ICU admission did not change significantly by exclusion, neither for the total group, nor for the three infection groups separately. In contrast, overall in-hospital mortality became lower (5.0% versus 10.6%, p = 0.023), but again without significant difference within the three infection groups, probably due to the low number of deceased patients.