Influenza A infections: predictors of disease severity

Abstract

Influenza affects approximately 10% of the world’s population annually. It is associated with high morbidity and mortality rates due to its propensity to progress to severe acute respiratory infection, leading to 10–40% of hospitalized patients needing intensive care. Characterizing the multifactorial predictors of poor prognosis is essential for developing strategies against this disease. This study aimed to identify predictors of disease severity in influenza A-infected (IFA-infected) patients and to propose a prognostic score. A retrospective cross-sectional study was conducted with 142 IFA-infected out- and inpatients treated at a tertiary hospital between 2010 and 2018. The viral subtypes, hemagglutinin mutations, viral load, IL-28B SNPs, and clinical risk factors were evaluated according to the patient’s ICU admission. Multivariate analysis identified the following risk factors for disease severity: neuromuscular diseases (OR = 7.02; 95% CI = 1.18–41.75; p = 0.032), cardiovascular diseases (OR = 5.47; 95% CI = 1.96–15.27; p = 0.001), subtype (H1N1) pdm09 infection (OR = 2.29; 95% CI = 1.02–5.15; p = 0.046), and viral load (OR = 1.43; 95% CI = 1.09–1.88; p = 0.009). The prognosis score for ICU admission is based on these predictors of severity presented and ROC curve AUC = 0.812 (p < 0.0001). Our results identified viral and host predictors of disease severity in IFA-infected patients, yielding a prognostic score that had a high performance in predicting the IFA patients’ ICU admission and better results than a viral load value alone. However, its implementation in health services needs to be validated in a broader population.

Supplementary Information

The online version contains supplementary material available at 10.1007/s42770-023-01186-w.

Introduction

Influenza is an acute respiratory viral infection with high transmissibility and global distribution, which presents seasonal epidemics, affects all social and age groups, and accounts for approximately 10% of the worldwide population annually. An influenza epidemic may substantially burden the economy worldwide due to high expenses on health services [1].

Influenza can cause upper respiratory tract infection (influenza-like illness) and lower respiratory tract involvement, leading to severe acute respiratory infection (SARI). Patients with chronic diseases, children, older adults, and pregnant women are at risk of developing severe illnesses. However, in 20% of the patients who progress to SARI, no clinical risk factors are found, suggesting that other viral and host factors may be associated with this outcome [2].

The innate immune response is an essential characteristic of the host, and it affects the progression of the disease. It has efficient mechanisms, such as cytokine induction, to control the most diverse types of viral infections [3]. Among them, lambda interferon (IFN-λ), whose family includes interleukin-28B (IL-28B) or interferon-λ3 (IFN-λ3), are important mediators in response to viral infection, especially in IFA infection [4]. IL-28B expression can be affected by single nucleotide polymorphisms (SNPs). Several SNPs have been studied, but their role in respiratory viral infections remains unclear. In contrast, studies on hepatitis C infections have already established the SNPs rs12979860 and rs8099917 as genetic factors related to spontaneous viral shedding and to ribavirin response [5].

Viral load (VL) and mutations in the viral genome, especially in the hemagglutinin protein, responsible for binding to the cell receptor, can also be considered possible predictors of disease severity. It is well established in the literature that mutations can confer an increase in viral fitness, virulence, and transmission and thus be associated with worse outcomes [6]. The role of VL IFA in the progression of infection is still unclear and, therefore, requires further investigations into the pathogenesis of the infection [7].

Vaccines and neuraminidase inhibitor antivirals are tools to reduce influenza impact. Immunization is available to high-risk groups, and its effectiveness depends on the vaccine coverage. The World Health Organization (WHO) has recommended the use of antivirals within 48 h of the onset of symptoms [1]. However, due to the high cost limiting access to this drug, the specific treatment is often not prescribed in primary health care services [8].

In this scenario, identifying novel human and viral biomarkers for disease severity may help track the cases and identify patients at higher risk for complications, therefore optimizing clinical management practices. Thus, this study aimed to identify predictors of disease severity in IFA-infected patients and to propose a prognostic score.

Material and methods

Study design

A retrospective cross-sectional study was conducted at Complexo Hospital de Clínicas/Universidade Federal do Paraná (CHC-UFPR), a tertiary academic hospital in southern Brazil. The Institutional Ethics Committee board approved the study (N#18714013.4.0000.0096).

Samples

The study was conducted with samples of convenience, which consisted of nasopharyngeal swabs or nasopharyngeal aspirates (NPA) collected from outpatients and inpatients admitted at CHC-UFPR between January 2010 and December 2018. Collection of respiratory samples from immunosuppressed patients is performed regardless of the time of onset of symptoms and immunocompetent patients; collection is performed within 3 days after the onset of symptoms. All clinical samples are processed and separated into aliquots, which are kept at − 80 °C for future repetitions. The genetic material after extraction is kept at − 20 °C until the tests are carried out, and then at − 80 °C. These clinical specimens were submitted for routine diagnostic testing and were positive for IFA infection. During the study, 4406 samples were analyzed to detect respiratory viruses at the Virology Laboratory of CHC-UFPR, out of which 153 (3.3%) samples were positive for IFA, and 142 were included in the study. Samples with insufficient volume or those that were not stored correctly were excluded. Clinical and epidemiological data were reviewed from the patients’ medical records.

Detection and molecular characterization of IFA

Viral RNA was extracted from 200 µL of clinical samples using the QIAmp Viral RNA mini kit (Qiagen, Hilden, Germany) following the manufacturer’s instructions. IFA subtyping was performed by reverse transcription-quantitative polymerase chain reaction (RT-qPCR) in the ViiA ™ 7 Real-Time PCR System (Applied Biosystems, Foster, CA, USA) with SuperScript™ III Platinum™ One-Step RT-qPCR Kit (Invitrogen, CA, USA) following the WHO protocol (2017), which simultaneously investigates influenza A and B, human adenovirus, respiratory syncytial virus, human rhinovirus, human metapneumovirus, group parainfluenza viruses, human bocavirus, human coronaviruses HKU1, 229E, OC43, and NL63 [9].

As described in previous studies, two viral strains that were co-circulating during the 1918 influenza A (H1N1) pandemic exhibited distinct receptor binding specificities related to mutations at positions 190 and 225 (H3 numbering) in the hemagglutinin (HA) gene. In this way, as Chen et al. (2010) [10], to assess polymorphisms at the receptor binding site in the H1N1pdm2009 subtype, the partial hemagglutinin segment (aa122-247) was sequenced using BigDye Terminator V. 3.1 Kit Cycle Sequencing (Applied Biosystems, Foster City, CA, USA) in a 3130 Genetic Analyzer (Applied Biosystems, Foster, CA, USA). The consensus sequences were obtained using the DNASTAR Lasergene SeqMan program v.7.0 (DNASTAR, Inc., Madison, WI), aligned with Clustal W in the Mega v.7.0 platform. Clinical association analyses were performed only for mutations that appeared in more than three patients.

Influenza A viral load

Absolute quantification of VL was performed using RT-qPCR as previously described by Pereira et al. (2021) [11]. The VL in copies/reaction (quantity) was converted into copies/mL using the formula: [Quantity/RT-qPCR volume reaction] * [sample elution volume (µL)/sample volume in the extraction (mL)]. Data were presented as Log₁₀ copies/mL.

IL-28B genotyping

To assess the host’s genetic factors, the IL-28B SNP rs12979860 and rs8099917 were characterized by RT-qPCR using the SNP TaqMan™ Genotyping Assays (Applied Biosystems, Foster City, CA, USA) following the manufacturer’s specifications. In rs12979860, the allele C > T, and in rs8099917, the allele T > G. The allele and genotype frequencies were calculated using HARDY–WEINBERG equilibrium.

Definitions

Outpatients: Individuals presenting influenza-like illnesses (ILI) during outpatient appointments. Inpatients: SARI patients from primary and secondary care health services. Risk factors: underlying diseases, pregnancy, age (≤ 2 and > 60 years old), immunosuppression (HIV-infected patients, hematological neoplasms, and use of systemic corticosteroids). Outcomes: primary: intensive care unit (ICU) admission; secondary: length of stay, mechanical ventilation support, and death. Predictors: virus subtype, hemagglutinin mutations, viral load, gender, ethnic group, IL-28B SNPs, and risk factors. Potential confounders: hematopoietic stem cell transplantation. Modifier effects: immunosuppression, co-infections, antiviral treatment, and sampling > 48 h of admission. Biases: use of convenience sample, lack of uniformity in the sample type, time of illness until sample collection, and data acquisition from medical records.

Data analysis

Statistical analyses were performed using the R version 3.6.1. Univariate analysis was performed using Fisher’s exact and Chi-squared tests for categorical variables and the Mann–Whitney and Kruskal–Wallis rank sum tests with Tukey’s multiple post hoc comparisons for continuous variables, as appropriate. Spearman’s correlation coefficient was used to assess the correlation between VL and length of stay, VL, and time between sample collection and hospital admission. A stepwise linear regression model was used to determine whether the VL was associated with patient characteristics. Multivariate logistic regression analysis was performed to assess covariates related to infection by (H1N1)pdm09 and ICU admission. Adjusted odds ratio (aOR) was calculated using the multivariate model with a stepwise selection of variables, with a cutoff point of p < 0.2. Pearson and residual deviance analyses were performed to test the fit model. Receiver operating characteristic (ROC) curve analysis was performed to assess the performance of the patients’ VL and the prognostic disease severity score in identifying probable ICU admission cases. The score was obtained from the aOR values of each disease severity predictor divided by a correction factor (the lowest aOR value among the predictors). All statistical tests were two-sided, with a significance level set at p < 0.05. A confidence interval (CI) of 95% was used to adjust the estimates.

Results

Of the 153 samples available, 142 (93%) were included in this study – 84 (59%) samples were nasopharyngeal aspirates and 58 (41%) were nasopharyngeal swabs. The study steps are outlined in the flowchart (Fig. 1).

Fig. 1.

Flow diagram of the study

Clinical-epidemiological characteristics

The median age was 22.8 years (IQR 3.4—53.5), 123 (87%) had risk factors, of which 98 (80%) had comorbidities, with hematopoietic stem cell transplantation (HSCT) and heart disease as the most common ones. Antiviral therapy was performed in 77% (n = 89) of SARIs and 63% (n = 90) of the total cases. From these, 85% (n = 76) started antiviral therapy after 48 h of admission to the hospital. SARI patients represented 82% (n = 116) of the sample, of which 59 (51%) needed ICU and 9 (15%) died. The univariate analyses showed that the patient profile in the ICU consisted of age over 60 years (p = 0.004) with cardiovascular diseases (p = 0.001) and HSCT (p = 0.001). In addition, the median length of stay was 3.5 times higher among patients who stayed in the ICU (median 11d; IQR 5.5–20; p < 0.0001) (Table 1).

Table 1.

Clinical-epidemiological characteristics of influenza A patients admitted to the CHC-UFPR between 2010 and 2018

	Total, n = 142 (%)	ICU, n = 59 (%)	non-ICU, n = 83 (%)	p value
Clinical samples, n (%)
Nasopharyngeal aspirates	84 (59)	34 (58)	50 (60)	0.863
Nasopharyngeal swabs	58 (41)	25 (42)	33 (40)
Median time between hospitalization and sample collection, days	2 (2–4)	2 (2–4)	2 (1–4)	0.635
Viral coinfection, n (%)	16 (11)	8 (13)	8 (10)	0.592
Ethnic group, n (%)
Caucasian	124 (87)	52 (88)	72 (87)	0.496
African descendant	18 (13)	7 (12)	11 (13)
Gender, male, n (%)	69 (49)	31 (52)	38 (46)	0.496
Age median, years (IQR)	23 (3.4–53.5)	38 (3.4–64)	21 (3.2–41)	0.197
Age group, n (%)
≤ 2 years	30 (21)	11 (19)	19 (23)	0.004
2–18 years	29 (20)	15 (25)	14 (17)
19–60 years	56 (40)	15 (25)	41 (49)
> 60 years	27 (19)	18 (31)	9 (11)
Risks factors, n (%)
Presenting any risk factors	123 (87)	52 (88)	71 (85)	0.804
Age group risk	58 (41)	30 (51)	28 (34)	0.059
Pregnancy	6 (4)	0	6 (7)	0.041
Presenting underlying diseases	98 (69)	42 (71)	56 (67)	0.714
Underlying diseases, n (%)
Asthma	21 (30)	12 (20)	9 (11)	0.211
Autoimmune	3 (2)	0	3 (4)	0.266
Cardiac disorders	29 (20)	19 (32)	10 (12)	0.001
Diabetes	6 (4)	5 (8)	1 (1)	0.082
HSCT	34 (24)	5 (8)	29 (35)	< 0.001
Neuromuscular disorders	8 (6)	6 (8)	2 (1)	0.067
Other lung disorders	20 (14)	13 (22)	7 (8)	0.445
Influenza vaccine, n (%)
Yes	34 (24)	16 (27)	18 (22)	-
Not described	108 (76)	43 (73)	65 (78)
Symptoms, n (%)
Cough	113 (80)	51 (86)	62 (75)	0.096
Dyspnea	94 (66)	51 (86)	43 (52)	< 0.0001
Fever	123 (87)	54 (91)	69 (83)	0.211
Myalgia	27 (19)	12 (20)	15 (18)	0.829
Sore throat	27 (19)	14 (24)	13 (15)	0.283
SpO2 ≤ 95%	95 (67)	54 (91)	41 (49)	< 0.0001
Oseltamivir therapy, n (%)	90 (63)	40 (68)	50 (60)	0.382
Antiviral until 48 h of hospital admission, n (%)	76 (85)	36 (61)	40 (48)	1.000
Disease severity, n (%)
ILI	26 (18)	0	26 (31)	< 0.0001
SARI	116 (82)	59 (100)	57 (68)
Respiratory support, n (%)	85 (60)	52 (88)	33 (40)	< 0.0001
None	57 (40)	7 (12)	50 (60)	< 0.0001
Oxygen supplementation	49 (35)	16 (27)	33 (40)
Mechanical ventilation	36 (25)	36 (61)	0
Median length of stay, days (IQR)	7 (2.0–13.5)	11 (5.5–20)	3 (0.0–7.5)	< 0.0001
Death, n (%)	9 (6)	9 (15)	0	-

ICU intensive care unit, HSCT hematopoietic stem cells transplantation, SpO2 ≤ 95% oxygen saturation ≤ 95%, ILI influenza-like illness, SARI severe acute respiratory infections. Values in bold indicate statistically significant

Influenza A viral load

Patients with worse outcomes had higher VL, ICU admission (p = 0.012), and death rates (p = 0.040). Patients with nasopharyngeal aspirate samples (p = 0.040) or with collection over 48 h (p = 0.22) also had a higher viral load. All differences between groups were greater than 0.5 Log₁₀.

Subjects were stratified into outpatients and inpatients with or without the need for ICU. The VL was significantly higher among the inpatient group (p = 0.033) compared to the outpatient group. A weak positive correlation was also found between VL and length of stay (r = 0.171; p = 0.042) (Fig. 2). In addition, an inversely proportional correlation between VL and the time elapsed between sample collection and hospital admission was found (r =  − 0.299; p = 0.001).

Fig. 2.

Viral load assessment in influenza A patients at the CHC-UFPR. a Viral load according to disease severity; b Spearman’s correlation between the viral load and the length of hospital stay of patients

Multiple linear regression analyses showed that the need for ICU (β =  + 0.63 Log₁₀ copies/mL; 95% CI = 0.07–1.18; p = 0.027) was an independent factor associated with VL increase, while samples collected 48 h after admission to the hospital were associated with a VL decrease (β =  − 0.64 Log₁₀ copies/mL; 95% CI =  − 1.25 – − 0.02, p = 0.042).

Virus subtyping

Figure 3 demonstrates the distribution of influenza A subtypes in the nine seasons evaluated. An alternating prevalence between (H1N1)pdm09 and H3N2 was observed, with a higher frequency of H3N2 infections (Fig. 3).

Fig. 3.

Distribution of influenza A cases treated at CHC-UFPR between 2010 and 2018 according to disease severity (primary axis) and subtype (secondary axis)

Univariate analysis showed that subjects infected with IFA (H1N1)pdm09 had more viral co-infections (p = 0.013), higher ICU hospitalization (p = 0.047), and longer hospital stay (p = 0.022), despite having fewer risk factors (p = 0.046) and higher antiviral use (p = 0.002) (Table 2).

Table 2.

Comparison of the clinical findings between H1N1(pdm09) and H3N2 influenza A subtypes

	(H1N1)pdm09, n = 53 (37%)	H3N2, n = 89 (63%)	p value	aOR (95% CI)	p
Risks factors, n (%)	42 (79)	81 (91)	0.046
1 risk factor	31 (58)	55 (62)	0.137	6.14 (1.68–22.46)	0.006
> 1 risk factor	11 (21)	26 (29)		7.70 (1.93–30.68)	0.004
None risk factor	11 (21)	8 (9)
Symptoms, n (%)
Cough	49 (92)	54 (61)	0.008	11.55 (2.26–59.05)	0.003
Dyspnea	44 (83)	58 (65)	0.066
Sore throat	14 (26)	13 (14)	0.160
Viral Coinfection, n (%)
	11 (21)	5 (6)	0.013	5.79 (1.53–21.84)	0.010
Oseltamivir therapy, n (%)
Yes	42 (79)	47 (53)	0.002
No	11 (21)	42 (47)
Disease severity, n (%)
Outpatient	7 (13)	19 (21)	0.047
Inpatient
ICU admission	29 (55)	30 (34)
Not ICU admission	17 (32)	40 (45)
Respiratory support strategies, n (%)
None	17 (32)	40 (45)	0.158
Oxygen supplementation	19 (36)	30 (34)
Mechanical ventilation	17 (32)	19 (21)
Length of stay, days (median, IQR)	9 (3.3–19)	5 (1.5–11)	0.022	0.98 (0.96–0.99)	0.048

The data included in the table are composed of variables with a p > 0.2 in the univariate analyses

ICU intensive care unit, aOR adjusted odds ratio. Values in bold indicate statistically significant

Multivariate analysis showed that IFA H3N2 infection was more likely to be present in patients with one or more risk factors (OR = 6.14; p = 0.006 and OR = 7.70; p = 0.004, respectively), without cough (OR = 11.55; p = 0.003), fewer viral co-infections (OR = 5.79; p = 0.010), and shorter hospital stay (OR = 0.98; p = 0.048) than H1N1(pdm09) infection.

Influenza A (H1N1)pdm09 hemagglutinin sequencing

In this study, IFA (H1N1)pdm09 infections were directly related to ICU admission and high length of hospital stay. Therefore, we investigated whether the presence of mutations in the hemagglutinin of this subtype would be associated with the modulation of VL and with worse outcomes. A total of 48 (92%) partial hemagglutinin segments (380 bp) were sequenced. Hemagglutinin 122-247aa region sequencing revealed 16 amino acid substitutions at 15 positions – 14 mutations at the Sa, Sb, Ca1, and Ca2 antigenic sites and two changes at an undesignated site (Fig. 4). No relationship was observed between the presence of mutation and disease severity (Supplementary Table 1).

Fig. 4.

The hemagglutinin region 122-247aa amino acidic alignment in (H1N1)pdm09 patients: (a) amino acidic sequence – *non-silent mutations (WebLogo 3 – http://weblogo.threeplusone.com/create.cgi); (b) amino acidic alignment by year (highlighter)

IL-28B genotyping

The evaluation of IL-28B SNPs in rs12979860 showed a frequency of 28% in the major allele (C) and 40% and 52% in the CT and TT genotypes, respectively. In the rs8099917, the major allele (T) frequency was 61%, and the frequencies of TG and GG genotypes were 48% and 15%, respectively. There was no association between genotypes or haplotypes of SNPs or VL with disease severity (Fig. 5).

Fig. 5.

Evaluation of IL-28B SNPs of IFA-infected patients at CHC-UFPR: a disease severity according to rs12979860 genotypes; b disease severity according to rs8099917; c genotypes VL according to rs12979860 genotype; d VL according to rs8099917 genotypes

Predictors of disease severity

Multivariate analysis identified that IFA (H1N1)pdm09 infection (OR = 2.29; 95% CI = 1.02–5.15; p = 0.046), high VL (OR = 1.43; 95% CI = 1.09–1.88; p = 0.009), cardiovascular disorders (OR = 5.47; 95% CI = 1.96–15.27; p = 0.001), and neuromuscular diseases (OR = 7.02; 95% CI = 1.18–41.75; p = 0.032) were independent factors associated with ICU admission. The presence of asthma was related to a higher frequency of invasive and non-invasive respiratory support (p = 0.06). Other pulmonary diseases were associated with ventilatory support (p = 0.014) and had longer hospital stays (12 to 6 days; p = 0.012). Among other lung diseases, chronic obstructive pulmonary disease (COPD) was the most frequent (n = 11/20). Antiviral treatment was not associated with the primary outcome. However, among patients who received treatment, it was observed an increase in the median length of stay (19 to 6.5 days; p < 0.0002) among those who started the antiviral after 48 h of hospital admission.

We have proposed a novel predictive prognostic index based on the adjusted odds ratio (aOR) obtained from multivariate analyses. The index comprises several factors, each assigned a specific point value: viral load (VL) greater than 6.45 Log₁₀ copies/mL (1 point), infection of subtype (H1N1)pdm09 (2 points), cardiovascular diseases (4 points), neuromuscular diseases (5 points), and asthma with the same viral load (1 point). The sum of these points results in a total score for each individual.

Using receiver operating characteristic (ROC) analysis, the index demonstrated strong predictive ability, with an area under the curve (AUC) of 0.812 (p < 0.0001). This indicates that the index is effective in distinguishing between different outcomes. Upon selecting a score cutoff of > 3, the model achieved an accuracy of 77.5%, with a positive predictive value (PPV) of 66.7% and a negative predictive value (NPV) of 91.1%. When using a higher cutoff score of > 4, the accuracy remained satisfactory at 72.5%, and the PPV and NPV values were 73% and 86.5%, respectively. It is essential to consider the prevalence of the disease when interpreting the results. For a disease prevalence of 20%, the calculated PPV and NPV values apply (Fig. 6).

Fig. 6.

Comparison between the ROC curves for IFA infection predictive prognosis using the viral load (dotted line) and score (solid line) as severity predictors. SCORE > 3 – sensibility: 66.67; specificity: 85.37. SCORE > 4 – sensibility: 40.00; specificity: 96.34

Discussion

In this cohort of IFA-infected patients, we identified that the VL, (H1N1)pdm09 infection, cardiac, neuromuscular, and lung diseases were factors associated with admission to ICU, extended hospital stay, need for respiratory support, or death. Reducing vaccine coverage and restricting the early use of antivirals contribute to the increase in severe cases of influenza. Thus, understanding the risk factors associated with SARI is essential for implementing early interventions to reduce the disease burden [9].

The participants’ characteristics reflected the tertiary hospital profile; most were hospitalized with SARI and had comorbidities, constituting a risk group for severe outcomes. The low median age group and the high rate of ICU admission were at odds with the main profile of severe cases of influenza in non-pandemic periods [12]. However, despite being a cohort of young patients with broad and rapid antiviral access, the considerable lethality of patients admitted to the ICU aligns with what has already been described in previous studies [13].

A positive correlation between VL and the length of hospital stay was observed, corroborating the findings from previous studies [14, 15]. Regarding the association between high VL and disease severity, Lee et al. (2013) [16] described an increase of 1.5 Log₁₀ copies/mL in hospitalized patients (p = 0.003). Li et al. (2010) [15] showed a higher VL in patients with SARI (p < 0.001). In addition, a direct relationship between VL and death was also found, similar to findings from Giannella et al. (2010) [17]. However, in contrast to our results, they did not find any difference in the initial quantification of VL, only in the clearance profile (p = 0.040).

Higher VL has also been demonstrated in IFA (H1N1)pdm09 infections, another predictor of the disease severity [17]. These patients had higher ICU admissions and more extended hospital stays despite having fewer risk factors and frequent oseltamivir use. This same profile was reported by Delgado-Sanz et al. (2020) [18] and other studies that evaluated the impact of this strain on worse outcomes even after controlling for age differences and comorbidities [19].

Moreover, Segaloff et al. (2019) [20] reported that (H1N1)pdm09 affects young adults more severely. Caini et al. (2018) [21] suggested that knowledge regarding this subtype is an essential tool for clinicians, as IFA (H1N1)pdm09 patients with need close monitoring due to the increased risk of unfavorable outcomes, probably due to mutations accumulated by this subtype [7].

In each seasonal period, influenza viruses accumulate mutations in their genome that can occur in essential epitopes, such as viral hemagglutinin, originating new strains. Such mutations can also result in increased viral fitness, greater virulence, and escape of the immune responses, impacting the outcome [22]. Although our samples have amino acid exchanges with previously reported viral antigenic properties modulating characteristics, no mutations were related to disease severity in this analysis [23].

As viral features have different impacts on the IFA infection outcomes according to the affected population, other factors, such as genetic, clinical, and demographic factors, must be evaluated.

Regarding immune response, the characterization of IL-28 SNPs has been extensively investigated in several viral infections. This gene is responsible for IFN-λ production that has antiviral activity and can be found in pulmonary and hepatic tissues during viral infection [4]. The occurrence of IL-28 SNPs can directly affect the production of interleukins, altering their functions and causing worse IFA infections [24]. In this study, we observed that major alleles were found less frequently in our infected patients than in the global population [25, 26], Americans [25, 26], and individuals with other viral infections, such as RSV [27]. This may suggest that patients without major alleles in these SNPs are more susceptible to IFA infection. For the rs12979860 region, this difference is more prominent (28% vs. 58–64%), while for the rs8099917 region, the difference was 61% vs. 72–87%. Giamberardino et al. (2022) [28] also observed a lower frequency of these alleles in a pediatric population hospitalized with RSV in the same region as our study. However, when comparing the results of the main SNP allele rs12979860 in our cohort, the frequency difference was lower than in the pediatric population with RSV (28% vs. 45%).

Regarding pathogenesis, our study found no association between the disease evolution and the presence of SNPs in IFA-infected patients, similar to other studies, such as those conducted in children with bronchiolitis caused by RSV [29] and adults infected with IFA [30]. However, other SNPs in IL-1β, IL-6, IL-10, IL-17, IFITM1, and IFITM3 are also involved in the viral infection pathogenesis [31], and further investigation is necessary to better understand the role of interleukins and their SNPs on the severity of IFA infection.

High-risk groups for the severe disease have been extensively reported [32]. However, in this cohort, cardiovascular disorders (p = 0.001) and neuromuscular diseases (p = 0.032) were the only independent factors associated with higher odds of being admitted to the ICU. These findings are aligned with other systematic reviews, which point to both diseases and obesity as factors associated with worse outcomes [12, 33].

Previously, our group reported an association between asthma and VL increase (+ 0.89 Log₁₀ copies/mL) [11]. Although an association with ICU admission was not found, these groups correlated with oxygen supplementation (p = 0.006) and higher length of stay (p = 0.012). These findings agree with previous studies that asthmatics are more likely to be hospitalized for IFA infection than people without asthma. However, they do not have an increased risk of ICU admission or death. In contrast, COPD is a risk factor for poor outcomes [12, 34].

The early use of antiviral was associated with a shorter hospital stay, but it did not impact disease severity (p < 0.0002), as previously described [35]. Similarly, no association was found between antiviral therapy’s worst outcomes or reduced VL. Antiviral use in influenza has presented discrepant results, and additional studies must evaluate its impact.

Several screening scores of patients with bacterial community-acquired pneumonia (CAP) are available, such as PSI, CURB-65, and SMRT-CO, but they are ineffective in viral pneumonia [36]. Therefore, recent studies are focusing on evaluating other severity scores in viral CAPs (qSOFA, SIRS score) [37, 38] or on developing a derived index of severity disease predictors [39].

However, the few proposed scores have shown an acceptable correlation only in patients at advanced stages of the disease. Furthermore, their applicability is exclusive to ICU patients; therefore, they use variables that focus on severe physiological dysfunction. In this study, we propose a prognostic disease severity score that performs well in identifying patients at high risk for severe disease, resolving late-onset antivirals, and improving clinical outcomes in patient management through optimized screening. The score performed very well, but its implementation in primary care and public health programs is still challenging. It depends on the availability of specific tests in primary health care and the quantification of VL in tertiary health services. Furthermore, despite having proved to be a potentially useful tool in this study, its implementation in public health needs to be validated in a broader population [40].

This study had some limitations: (i) Most patients tend to undergo late hospitalization in this tertiary hospital. The hospital mainly receives patients coming from basic health units or emergency services, which delays their referral; (ii) there was difficulty in standardizing the date of sample collection after admission. Previous studies [15, 16] also reported a negative correlation between collection time after symptom onset or hospitalization and VL; (iii) although previous analyses did not show significant differences in the VL between the swabs and NPA [41, 42], the use of two types of samples in the quantification of VL had some limitations, which may lead to the underestimation of the results of viral quantification. This fundamental criterion must be normalized for VL’s future inclusion in diagnostic procedures. However, these findings reinforce the importance of studying factors associated with higher risk of adverse outcomes in individuals infected with influenza, allowing prompt interventions and, consequently, protecting these individuals.

Overall, our proposed predictive prognostic index shows promising potential in predicting outcomes in this context and could be a valuable tool for risk stratification and decision-making in clinical settings. Further validation and evaluation in diverse populations are warranted to fully establish its utility and generalizability.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

Pereira LA: data collection, data analysis and interpretation, drafting the article, final approval. Lapinscki BA: data analysis and interpretation, drafting the article, critical revision, final approval. Santos JS: critical revision and final approval. Debur MC: critical revision, final approval. Petterle RR: data analysis. Nogueira MB: critical revision, final approval. Vidal LRR: data analysis and interpretation, critical revision, final approval. De Almeida SM: critical revision, final approval. Raboni SM: conception, data analysis and interpretation, drafting the article, critical revision, final approval, funding acquisition. All authors approved the final manuscript.

Funding

This work was supported by the Fundação Araucária. Project number: 059/2017.

Data availability

The data that support the findings of this study are available on request from the corresponding author.

Declarations

Conflicts of interest

The authors declare no competing interests.