Signa Vitae. 2021; 17(5): 18-29. doi: 10.22514/sv.2021.148
Systematic Review

Prediction models for prognosis of influenza: a systematic review and critical appraisal

Yao Sun1,2,†, Yiwu Zhou1,†, Shu Zhang1,*,

1Department of Emergency Medicine, Emergency Medical Laboratory, West China Hospital, Sichuan University, 610041 Chengdu, Sichuan, China

2West China Medical School, West China Hospital, Sichuan University, 610041 Chengdu, Sichuan, China

*Corresponding Author(s):zhangs@wchscu.cn (Shu Zhang)

† These authors contributed equally.

† These authors contributed equally.

History Submitted: 18 July 2021 | Accepted: 23 August 2021 | Published: 08 September 2021
Copyright:  ©2021  The Author(s). Published by MRE Press.
This is an open access article under the CC BY 4.0 license (https://creativecommons.org/licenses/by/4.0/).

Collapse table of contents

Abstract

The influenza epidemic has become an important public health issue throughout the world. Early recognition of potentially terrible outcomes is important in the emergency department (ED). Efficient prognosis of the disease is conducive to reducing the financial burden and providing appropriate care for patients. Prediction models containing several features to estimate the risk of patients with confirmed infection could help clinicians give appropriate treatment when health care resources are limited. We conducted a literature review of studies about influenza published until June 2021 and updated the literature during the creation process. We researched PubMed, Web of Science, and Google Scholar databases to collect articles in English relevant to influenza between Jan 1, 1900, and Dec 30, 2020. The terms used for the search were “influenza”, “diagnostic”, “prognostic”, “prediction”, “score”, “artificial intelligence”, and so on. If the study involved animals, children, pregnant women or the study type was pragmatic and explanatory clinical trial, guideline, protocol, letter, a case report was also excluded. The GRACE checklist in our study was used to assess the 34 studies for quality. Thirty-four articles were included in the review, and relevant data were extracted from the risk prognosis model. Cardiovascular disease and central nervous symptoms play an important role in prognostic models of influenza. In addition, some commonly used scoring systems can also play a certain role in evaluation. This systematic review compared different types of models for predicting the prognosis of influenza infection, informing us of risk factors for the predictive model in predicting the prognosis of influenza in the early stage. The articles were limited to retrospective observational studies, sample size, time limitation, incomplete data, imbalanced prognosis treatment, and so on.

Keywords:Prediction models;Prognosis;Influenza;Review;Critical appraisal
PDF(588.02 kB)|EndNote (RIS)|BibTeX|RefMan|RefWorks

Cite this article

Yao Sun, Yiwu Zhou, Shu Zhang. Prediction models for prognosis of influenza: a systematic review and critical appraisal. Signa Vitae. 2021; 17(5): 18-29. doi: 10.22514/sv.2021.148

1. Introduction

Community influenza activity has long been a serious threat to global public health [1, 2, 3, 4, 5]. During seasonal epidemics, the number of severe cases registered worldwide is 3 to 5 million, and the number of lethal cases is 250,000–500,000 [6, 7]. The prognosis of influenza patients ranges from asymptomatic infection to severe pneumonia and even acute respiratory distress syndrome (ARDS) and multiple organ dysfunction syndrome [1, 8, 9, 10]. Influenza progresses rapidly, usually leading to morbidity and mortality within days [11, 12]. In addition, influenza will increase the mortality rate of elderly patients or patients with underlying diseases [5, 13, 14, 15, 16, 17, 18]. The evolution and reassortment of influenza viruses are the main power that would generate novel strains, against which the humans have no or little immunity [19, 20].

In the emergency department, it is important to identify flu patients with poor prognosis early and to be alert and pay attention to them [11, 21]. To reduce the burden on the health care system and provide patients with the best care, it is necessary to effectively predict the prognosis of the disease and carry out early intervention [1, 22, 23]. In clinical medical research, a large number of predictive models that combine multiple variables or characteristics have been developed to estimate the risk of people being infected or the risk of adverse consequences after infection, which can help medical staff treat patients when they allocate limited medical resources [24, 25, 26, 27, 28]. Models ranging from regressive scoring systems to advanced machine learning have been proposed and released in response to calls for rapid public sharing of relevant influenza research results to inform public health responses and help save lives [3, 9, 29, 30, 31]. However, the models developed using machine learning are mostly basic research on influenza virus infection, and do not include clinical evaluation content [32].

To date, there has not been a systematic review or meta-analysis of the prognostic model of influenza infection. We aim to systematically review and critically evaluate the currently available influenza prognostic prediction models.

2. Materials and methods

This study evaluates the predictive power of prognostic models of influenza patients in different studies.

2.1 Data sources and search strategy

First, the databases PubMed, Web of Science and Google Scholar were searched to collect and identify articles in English that were relevant to influenza between Jan 1, 1900, and Dec 30, 2020. Relevant searches were also carried out during writing. The key terms used for the search were “influenza”, “human influenzas”, “Flu”, “Grippe”, “diagnostic”, “prognostic”, “prediction”, “prediction model”, “regression”, “score”, “artificial intelligence”, “algorithm”, “deep learning”, “machine learning”, “CURB-65”, and “PSI”. We made various combinations of these words. We also included documents in the references that met the requirements for further inspection.

2.2 Study selection and data extraction

We included all English language full text articles that described retrospective and prospective observational studies, and randomized controlled trials. Inclusion criteria for studies were based on existing standards for diagnosis of influenza from research. If infection is suspected, a nasopharyngeal tract sample for conventional influenza RT-PCR should always be obtained. Antigen testing and direct or indirect antibody staining tests should only be used in settings lacking the more sensitive molecular assays. Participants included in the articles should be diagnosed with the infection of at least one type of influenza. The study should enroll adult patients (≥16 years) with in-hospital confirmed influenza. We chose the evaluated mortality as an outcome of interest (in-hospital, 30-day, or 90-day mortality) in our study. We did not include studies involving animals, children, or pregnant women, or those not written in English. Pragmatic and explanatory clinical trials, guidelines, protocols, letters, case reports, or case series were also excluded.

First, the titles and abstracts we drew were screened and selected by two authors (YS and YWZ). To ensure that information was accurate and determine final eligibility, all reviewers (YS, YWZ, and SZ) screened full texts of likely articles. The authors (YS and YWZ) further selected articles that were relevant to the prognosis of influenza. Meanwhile, we (YS, YWZ, and SZ) identified the risk factors related to influenza infection. We all worked independently, and team members all contributed to review and data extraction to avoid discrepancies and questions, reaching a consensus. If there was any disagreement, the third one would join in and give a decision.

We imported all searched articles into the literature management software Endnote 9.2 (Thomson Reuters, New York, NY, USA) and excluded all duplicate articles by comparing the title, authors, publication year, and name of the journal. Once we determined the articles, data was extracted and transferred to standardized form by one team member using Microsoft Office Excel and Word 2016 (Microsoft, Redmond, WA, USA), and the other team member verified the accuracy. The form was discussed and designed by all reviewers. The form-filling criteria were also determined by three creators.

This systematic review has applied and passed the international platform of registered systematic review and meta-analysis protocols (INPLASY 202120047) on February 16, 2021 (doi:10.37766/inplasy2021.2.0047), hoping that the whole process will be scientific and concrete.

2.3 Quality assessment

The process of study selection and data extraction in whole is presented in the flow chart following the PRISMA principle. We selected the GRACE (Good Research for Comparative Effectiveness) guidelines for rating the quality of studies [33]. Quality ratings were assigned by two separate raters (YS and YWZ) for each publication using the GRACE checklist, which was published by the Academy of Managed Care Pharmacy. It comprises 11 items, split into data and methods. At the same time, we used the GRADE as an auxiliary evaluation method, which is endorsed by the World Health Organization. the overall quality of evidence was graded as high, moderate, low, or very low, according to the Grading of Recommendations Assessment, Development and Evaluation guidelines. Both tools showed consistency. Any conflict was resolved by consensus. Every team member participated in improvement of the GRACE checklist.

3. Results

3.1 Study search result and selection

The searches in PubMed, Medline, Web of Science and Google Scholar initially resulted in 4672 potential eligible articles. We also collected 122 articles through references and other related methods, initially retrieving 4778 articles after using the document manager EndNote to remove the same ones. After reading the titles and abstracts, 268 remained for further full-text screening. Finally, 28 (10%) of 268 met the inclusion criteria for the review, and 240 (90%) were excluded because of the following reasons: 8 (3%) for no outcome, 17 (7%) for pregnancy, 13 (6%) for children, 25 (10%) for the study type doesn’t match, and 177 (74%) for other reason that didn’t meet the inclusion criteria. We once again retrieved 6 additional articles related by browsing the full text and related information (Fig. 1) [34, 35, 36, 37, 38, 39].

Flow diagram of study selection process.

Fig. 1.Flow diagram of study selection process.

Carefully reading the original text, we extracted the year of article, types of article, geographic region, data collection, number of centers and patients, average age, and final outcome. We collected the overall evaluation indicator AUCs for the model provided by each article (Appendix Table 3, Ref. [1, 3, 6, 9, 11, 24, 25, 26, 28, 29, 30, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56]).

Eighteen (53%) items focused on 2009 and 2010 [26, 29, 35, 37, 38, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50]. Only 4 (12%) collected patient data earlier than 2009 [24, 34, 51, 52]. Twelve (35%) articles belonged to a multicenter study based on the prevalence of influenza [26, 34, 36, 40, 41, 42, 43, 45, 46, 47, 48, 51, 53]. However, only 5 (15%) were prospective studies [6, 36, 41, 43, 47, 50, 54].

3.2 The risk factors

For different models included in the studies, we reviewed the risk factors related to the prognosis and outcome of influenza. Then, we summarized different risk factors and displayed them in the form of a table. A total of 42 relevant risk factors were calculated in 34 articles, and each factor had a different proportion (Table 1).

Table 1.The risk factors for the prognosis of influenza.
CategoryContentTimesPercentage (Times) aTotal times
Population data and medical historyAge50.0625
Gender10.01
Smoking10.01
BMI10.01
Cardiovascular disease70.09
Atrial fibrillation10.01
Kidney disease30.04
Malignant disease30.04
Diabetes mellitus10.01
Metabolic syndrome10.01
Number of comorbidities10.01
Clinical symptomsDyspnea40.0512
CNS symptoms b70.09
Thoracic pain10.01
Vital signsBlood pressure10.013
Respiratory rate10.01
Dehydration10.01
Laboratory testCRP20.0312
LDH20.03
ALT10.01
D-dimer10.01
Lymphopenia10.01
Leucocyte count10.01
Bandemia10.01
Thrombocytopenia10.01
Hypoalbuminemia10.01
Nosocomial infections10.01
ImagingPneumonia20.034
Abnormal CR or CT20.03
Treatment cNeuraminidase inhibitors10.012
Statin use10.01
Score systemShock index20.0319
PaO2/FiO2 ratio40.05
qSOFA30.04
PSI20.03
APACHE II20.03
iPIT10.01
GID10.01
STSS10.01
SIRS10.01
OHPIP10.01
ILI-score10.01
Total (34 articles)42771.0077

a: The number keeps two decimal places; b: Central nervous system; c: Reduced mortality.

In regard to population and medical history, age and cardiovascular disease seem to have a relatively large impact on the prognosis of influenza. Central nervous system (CNS) symptoms play an important role in clinical manifestations. Blood pressure, respiratory rate, and dehydration of the body state have equal shares and cannot be ignored. In laboratory tests, inflammation, abnormal liver function, and a reduction in the number of various types of cells all affect the outcome of patients. In particular, nosocomial infections might be one of the risk factors that we should consider in the future. In addition, it is also necessary to consider whether upper respiratory tract and other viral infections would affect the outcome of the patient. Many types of viral infections can cause thrombocytopenia [57]. The pathogenesis includes the formation of immune complexes, changing the structure of platelet membrane glycoproteins, activating the complement system, forming nuclear virus inclusion bodies and enhancing the activity of the mononuclear-macrophage system. Thrombocytopenia is also one of the prognostic indicators of influenza virus infection according to our review. If there are obvious lung imaging abnormalities on imaging, it means that there has been an organic disease, and the virus has affected the respiratory system. In approximately 4 times the models mentioned, pneumonia imaging and abnormal CR or CT should be risk factors for influenza infection. Many common severe respiratory disease or self-created scoring systems showed good guidance, and the PaO2/FiO2 ratio, which shows the oxygenation of blood gas in the lungs, performed the best. Common interventions such as oral antiviral drugs, even if taking statin drugs, could improve the poor prognosis of patients.

3.3 Quality assessment

The GRACE checklist in our study was used to assess the 16 studies for quality (Table 2). The 11-item checklist assesses data attributes and methods (Items D1–6 and M1–5, respectively) [33]. Items were classified as sufficient or insufficient in accordance with a qualitative judgment by the assessors. Most of the included studies were deemed to be at low risk of bias. Studies rarely were rated as high risk of bias mainly according to the GRACE checklist.

Table 2.GRACE assessment scores by item.
Checklist itemTotal (n)% of total
D1. Were treatment and/or important details of treatment exposure adequately recorded for the study purpose in the data source(s)?1956
D2. Were the primary outcomes adequately recorded for the study purpose?2162
D3. Was the primary clinical outcome(s) measured objectively rather than subject to clinical judgment?2368
D4. Were primary outcomes validated, adjudicated, or otherwise known to be valid in a similar population?1132
D5. Was the primary outcome(s) measured or identified in an equivalent manner for the treatment/intervention group and the comparison group?1647
D6. Were important covariates that may be known confounders or effect modifiers available and recorded?824
M1. Was the study (or analysis) population restricted to new initiators of treatment or those starting a new course of treatment?1338
M2. If one or more comparison groups were used, were they concurrent comparators? If not, did the authors justify the use of historical comparison groups?1853
M3. Were important confounding and effect-modifying variables taken into account in the design and/or analysis?1647
M4. Is the classification of exposed and unexposed person-time free of immortal time bias?1132
M5. Were any meaningful analyses conducted to test the key assumptions on which the primary results are based?1544

4. Discussion

This systematic review demonstrates that a large number of influenza infection risk prediction models have been developed in the past few decades. We identified 34 studies that reported 42 risk factors, which showed great heterogeneity in the selection and definition of predictive factors. Prognostic models of influenza infection are all available, and they all seem to show good prognostic performance in the study. However, some models have a high risk of bias, and the sample size is too small. In addition, this is because the selection of patients in the group is not representative and the model is overfitting [6, 9, 11, 40]. Therefore, the performance estimates of some models are likely be misleading or optimistic [11, 25, 34, 42]. We have proven that some of the definitions and assessments of risk factors in the research are quite different, which may appear to be the same on the surface, and in many cases, the definitions and assessment methods given in the research lack detailed descriptions [1, 9, 11, 28, 29, 40, 42, 55]. However, even if the definitions of risk factors are the same in models, there are still many differences in detailed descriptions. An analysis of the evaluation methods used in the research shows that the actual standards for performance measurement verification and reporting are not sufficiently optimized. Only a few studies have been externally validated, and some of them are prospective studies [6, 41, 50]. However, neither internal verification nor time verification can detect the general applicability of the model because it requires the use of data from different sources for external verification [58, 59, 60].

When developing a model, the target population must also be carefully described to evaluate the performance of the development or verification model, which determines the applicable population of the model when testing and using the model. However, studies included in our systematic review tend to overlook the full description of the research population, which makes users of these models doubt the applicable population of the model. Models developed in different regions and groups may not be suitable for another group and region, especially risk factors that may be different. The clinical application performance of the model in different regions and populations is a particularly valuable indicator for judging its ability [59, 61]. Therefore, it is very important to verify the risk prediction model externally to verify the portability and universality of the model to its region and population cohort [62]. For the comparison of different prediction models, external verification is the best way to determine the predictive ability of the model in independent datasets, although only a few models have been externally verified [60, 63, 64]. A detailed description of the study population can also help us understand the variability of the results of the observed studies, such as mortality related to influenza infection. The variability of prediction results poses important challenges to many prediction models. Data from individual participants in the health care system in different regions may help to better understand the universality and practicability of prediction models in different regions and populations rather than being limited to a certain region and population for development and prediction [65]. This research method can greatly improve the universal applicability and robustness of the prediction model.

As Table 2 illustrates, there are various combinations of risk factors in all risk factor groups, but it is still possible to find a few frequent factor combinations. The most commonly included risk factors are patients with a comorbidity of cardiovascular disease and CNS symptoms on admission, which were used in seven of the 28 models. Both risk factors were included in most models for which they were considered, consistent with previously identified positive associations. This systematic review determined that comorbidities such as cardiovascular diseases play a key role in the prognosis of influenza-infected patients. According to reports, cardiovascular disease is one of the most important independent risk factors for many viral infections. Influenza virus infection can damage the function of multiple organs, including the lungs, liver, heart and kidneys. Studies have demonstrated that angiotensin-converting enzyme 2 (ACE2) may be the functional receptor for influenza virus to enter human cells. Influenza viruses may increase pulmonary vascular permeability and cause acute lung injury by downregulating ACE2 expression and increasing angiotensin II levels [66, 67, 68]. However, the specific mechanism of infection is still unclear, and more studies are needed to confirm that influenza virus infection in patients with cardiovascular disease will accelerate this process [69]. CNS symptoms have been proposed as a clinical sign in prediction models for the prognosis of influenza infection. Previous models have shown that age >65 is an important independent prognostic factor in patients with influenza infection.

From the results of various study reports, it can be seen that there are certain differences in the verification performance of each model. There are also some studies that have not reported any indicators of model performance, although quantitative indicators of model performance are important for evaluating model performance. For various types of performance measurement indicators, the proportion of studies reporting such measures is higher in the subgroups of newer studies than in the subgroups of older studies. For the risk prediction of infected individuals, the discriminative ability of the model is the most basic and most important attribute of the prediction model. Approximately 80% of studies report the discriminative ability of the model. The main indicator is the AUC or c index, but the discriminative ability of many models cannot be effectively applied in clinical practice [1, 3, 6, 28, 29, 30, 41, 52, 55].

However, there are several limitations. We only included studies published in English and did not search gray literature, but the missing models due to this are limited in usage and usually of relatively low quality. Most of the studies are retrospective observational studies, small sample sizes, incomplete data, and imbalanced prognostic treatment may all have an impact on predictive performance. These models may represent local practices, but due to the low proportion of patients from international centers, their generality outside the region is limited. Demonstration of the wider applicability of the models will require multiple studies aiming to externally validate them in different cohorts of patients.

5. Conclusions

The prognostic models for patients diagnosed with influenza infections seem to show good to excellent discriminating performance and are usable. However, some models have a higher risk of bias, mainly due to the different criteria for including patients and the definition of risk factors. Therefore, the predictive power of some models may be optimistic and misleading. Future research needs to share data and expertise used to develop, validate and update predictive models related to influenza infection.

Abbreviations

ED, emergency department; IPPV, invasive positive pressure ventilation ; CXR, chest X-ray; ICU, intensive care unit; PSI, pneumonia severity index; CRP, C-reactive protein; WBC, white blood cell; PMEWS, the Pandemic Medical Early Warning Score; qSOFA, ‘quick’ sequential organ failure assessment scores; APACHE II, Acute Physiology and Chronic Health Evaluation; ICNARC, Intensive Care National Audit & Research Centre; CURB-65, a score for pneumonia severity; NA, Not Available; IQR, interquartile range; BMI, body mass indes; LDH, lactate dehydrogenase; ALT, alanine transaminase; AUC, area under curve; iPIT, Influenza Pandemic ICU Triage; GID, Geriatric influenza death; STSS, the Simple Triage Scoring System; SIRS, systemic inflammatory response syndrome; OHPIP, Ontario Health Plan for an Influenza Pandemic; ILI-score, the influenza-like illness score.

Author contributions

Conception and design—YS, YZ, SZ. Collection and assembly of data—YS, YZ. Data analysis and interpretation—YS, YZ, SZ. Manuscript preparation—YS. Manuscript proofing and final approval of the manuscript—all authors.

Ethics approval and consent to participate

Not applicable.

Acknowledgment

We are grateful to the numerous individuals who participated in this study.

Funding

This study was funded by the Applied Basic Research Project of the Science and Technology Department of Sichuan Province (Process No. 2017JY0334).

Conflict of interest

The authors declare no conflict of interest.

Consent for publication

Written informed consent for publication was obtained from all participants.

Appendix

See Table 3.

Table 3.Details of the studies on the prognosis models of influenza.
SourceStudy designCountryStudy periodPatient of data collectionNO. of study sitesNO. of patientsAgesMortality definitionAUCEvidence qualitya
Cvetanovska M, et al. [6], 2016ProspectiveMacedonia2012–2015Age, sex, living place, vaccine, use of osaltamivir, intensive, mechanic ventilation, comorbid conditions, cardiovascular disease, SAPS 2 score18754.77 ± 17.3ICU0.755Moderate
Dead
Fujikura Y, et al. [40], 2014RetrospectiveJapan2009–2010Age, sex, and comorbid conditions, IPPV use, treatment options and outcome data2491346NAICU0.82Low
Dead
Capelastegui A, et al. [41], 2012ProspectiveSpain2009–2010Sociodemographic characteristics, pre-existing medical conditions, vaccinations, toxic habits, previous medications, exposure to social environments, measures to prevent influenza3661848.60 (SD 15.7)In-hospital0.77High
Tai H C, et al. [9], 2019RetrospectiveTaiwan2010–2015Vital signs, demographic characteristics, influenza subtype, laboratory data, past medical history, admission, and 30-day mortality data140979.5 ± 8.3In-hospital0.77Low
Schoen K, et al. [11], 2019RetrospectiveBrazil2016Clinical, laboratorial and radiological data116043ICUNALow
Chu S E, et al. [1], 2020RetrospectiveTaiwan2010–2016Vital signs, blood tests, image reports, diagnosis, treatments, and daily medical records of doctors and nursing staff1356148.08 ± 19.51In-hospital0.861Low
Chung J Y, et al. [28], 2018RetrospectiveTaiwan2010–2015Demographic characteristics, vital signs, past history, laboratory data, complications, and outcomes140979.5 ± 8.330-day dead0.861Low
Adeniji K A, et al. [29], 2011RetrospectiveUK2009–2010Demographic data, comorbidity, CXR, ventilatory support, level of care, days, mortality and the physiological and laboratory components16240.52In-hospital0.88Low
Choi W I, et al. [42], 2011RetrospectiveSouth Korea2009–2010Age, sex, intensive, mechanical ventilation, antiviral and antibacterial agents1426948 (Rang 15–93)In-hospitalNALow
Shi S J, et al. [55], 2017RetrospectiveChina2009–2014Characteristics, comorbidities, laboratory and image, vital signs117055.4 ± 17.7In-hospital0.945Low
Cheung W, et al. [43], 2012ProspectiveNew Zealand; Australia2009–2010Ontario Health Plan for an Influenza Pandemic (OHPIP) triage protocols8805NAICUNAHigh
Chang S H, et al. [3], 2019RetrospectiveTaiwan2010–2015Demographic data, vital signs, past histories, influenza subtypes, treatment outcomes, radiographic findings, readmission rates and qSOFA140979.5 ± 8.330-days dead0.81Low
Morton B, et al. [25], 2017RetrospectiveUK2010–2011Demographics, comorbidities, physiological observations, clinical laboratory tests, arterial blood gases and oxygen saturations186NANA0.88Low
Pawelka E, et al. [56], 2018RetrospectiveAustria2017–2018Demographic data, laboratory results, symptoms, treatment and underlying medical conditions139675.5 (Rang 63–84)In-hospitalNALow
90-day dead
Chung J Y, et al. [30], 2019RetrospectiveTaiwan2010–2015Demographic data, vital signs, shock index, past histories, subtypes and outcomes140979.5 ± 8.330-day dead0.62Low
Ho Y C, et al. [24], 2009RetrospectiveTaiwan2001–2007Demographics, symptoms, hospitalization, the presence of sepsis, severe sepsis, laboratory values, virus isolation, antiviral treatment and vaccination122552 (Rang 17–89)In-hospitalNALow
Muller M P, et al. [51], 2010RetrospectiveCanada2005–2007Demographic, clinical, laboratory and radiographic data2561776 (Rang 64–83)ICU0.8Low
In-hospital
Rodriguez-Noriega E, et al. [44], 2010RetrospectiveMexico2009Demographics, signs and symptoms, history of health care utilization, chronic medical conditions, laboratory and radiology findings1184029 (Rang 22–41)NANALow
Zhang P J, et al. [45], 2013RetrospectiveChina2009Demographic information, underlying conditions, vaccination status, medication, complications and outcomes426215134.0 (IQR 24.1–50.6)In-hospitalNALow
Cho W H, et al. [26], 2011RetrospectiveSouth Korea2009Demographic data, PSI, CURB65, risk factors, time to first dose of antiviral medication, routine laboratory data, clinical outcome and radiological characteristics23746.1 ± 17.3In-hospitalNALow
Riquelme R, et al. [46], 2011RetrospectiveUSA2009Demographic data, history and physical findings, comorbidities, laboratory characteristics2225043.3In-hospital0.78Low
Pereira J M, et al. [47], 2012ProspectiveSpain2009Demographics, comorbid conditions, physiological status and organ supports3326542 ± 16.1ICU0.73High
Bjarnason A, et al. [54], 2012ProspectiveIceland2008–2009Sputum, blood cultures, PSI, CURB-65 and APACHE II scores111444.0 (95% CI 37.1–50.9)ICUNALow
Commons R J, et al. [48], 2012RetrospectiveAustralia2009Clinical characteristics, comorbid conditions, potential infection, symptoms, treatment, PSI, CURB-65 and SMRT-CO severity scores711242 (Rang 15–79)ICU0.83Low
Kiliç H, et al. [49], 2015RetrospectiveTurkey2009Age, sex, duration, fever, CRP, sedimentation rate, WBC counts, platelet counts, monocytosis, leukocytosis, D-dimer levels, CURB-65 scores, comorbid illnesses and radiological findings17556.5 (Rang 17–85)ICUNALow
In-hospital
Challen K, et al. [52], 2007RetrospectiveUK2005Clinical characteristics, PMEWS, CURB-651144NAIn-hospital0.944Low
Brandão-Neto R A, et al. [50], 2012ProspectiveSpain2009Underlying medical conditions, laboratory tests, radiographic findings, pneumonia severity, treatment course, in-hospital death, ICU admission, invasive mechanical ventilation, vasopressor use, renal failure at the admission and during the follow-up15343ICUNAModerate
In-hospital
Atamna A, et al. [39], 2021RetrospectiveIsrael2017–2018baseline demographics, BMI (kg/m2), Comorbidities, hypertension, ischemic heart disease, congestive heart disease, chronic obstructive pulmonary disease, chronic kidney disease, liver disease, malignancies, and organ transplantations151274 (Rang 62–83)30-day deadNALow
Louie J K, et al. [35], 2011RetrospectiveUSA2009Height, weight, demographic characteristics, clinical presentation and course, comorbid conditions, and laboratory and radiographic findings, hospital admission, infectious disease, radiographic and microbiologic studies, transfers, and hospital discharges onto a standardized case report form153446 (Rang 20–92)DeadNALow
Demirjian S G, et al. [37], 2011RetrospectiveUSA2009Demographic, clinical, and outcome data18946 without AKI; 49 with AKIDeadNALow
Wong C M, et al. [34], 2013RetrospectiveChina1998–2001Baseline data, lifestyle habits (smoking history, exercise frequency and alcohol drinking), socioeconomic status (housing type, education and monthly expenditure) of all the subjects1866820NADeadNALow
Lopez-delgado J C, et al. [36], 2013ProspectiveSpain2009–2011Demographic, clinical, and outcome data211449.2 ± 14DeadNAModerate
Rowan K M, et al. [38], 2016RetrospectiveIran2009Demographics, occupation, ethnicity, physical examination, blood pressure, height and weight155NAICUNALow
Hypoxemia
Rowan K M, et al. [53], 2010RetrospectiveUK2007–2010Age, sex, source of admission to the critical care unit, APACHE II, ICNARC physiology score, CURB-6522156258.8ICUNALow
In-hospital

References

Chu S, Seak C, Su T, Chaou C, Tseng H, Li C. Prognostic accuracy of SIRS criteria and qSOFA score for in-hospital mortality among influenza patients in the emergency department. BMC Infectious Diseases. 2020; 20: 385.

[Google Scholar]

Nishiura H. Case fatality ratio of pandemic influenza. The Lancet Infectious Diseases. 2010; 10: 443–444.

[Google Scholar]

Chang S, Yeh C, Chen Y, Hsu C, Chen J, Chen W, et al. Quick-SOFA score to predict mortality among geriatric patients with influenza in the emergency department. Medicine. 2019; 98: e15966.

[Google Scholar]

Nguyen JL, Yang W, Ito K, Matte TD, Shaman J, Kinney PL. Seasonal Influenza Infections and Cardiovascular Disease Mortality. JAMA Cardiology. 2016; 1: 274–281.

[Google Scholar]

Cox NJ, Subbarao K. Global epidemiology of influenza: past and present. Annual Review of Medicine. 2000; 51: 407–421.

[Google Scholar]

Cvetanovska M, Milenkovik Z, Uroshevik VK, Demiri I, Cvetanovski V. Factors Associated with Lethal Outcome in Patients with Severe Form of Influenza. Prilozi (Makedonska Akademija Na Naukite i Umetnostite. Oddelenie Za Medicinski Nauki). 2016; 37: 63–72.

[Google Scholar]

Wong JY, Kelly H, Ip DKM, Wu JT, Leung GM, Cowling BJ. Case fatality risk of influenza a (H1N1pdm09): a systematic review. Epidemiology. 2013; 24: 830–841.

[Google Scholar]

Ergönül Ö, Alan S, Ak Ö, Sargın F, Kantürk A, Gündüz A, et al. Predictors of fatality in pandemic influenza a (H1N1) virus infection among adults. BMC Infectious Diseases. 2014; 14: 317.

[Google Scholar]

Tai HC, Yeh C, Chen Y, Hsu C, Chen J, Chen W, et al. Utilization of systemic inflammatory response syndrome criteria in predicting mortality among geriatric patients with influenza in the emergency department. BMC Infectious Diseases. 2019; 19: 639.

[Google Scholar]

Rothberg MB, Haessler SD, Brown RB. Complications of viral influenza. The American Journal of Medicine. 2008; 121: 258–264.

[Google Scholar]

Schoen K, Horvat N, Guerreiro NFC, de Castro I, de Giassi KS. Spectrum of clinical and radiographic findings in patients with diagnosis of H1N1 and correlation with clinical severity. BMC Infectious Diseases. 2019; 19: 964.

[Google Scholar]

Shah NS, Greenberg JA, McNulty MC, Gregg KS, Riddell J, Mangino JE, et al. Severe Influenza in 33 us Hospitals, 2013-2014: Complications and Risk Factors for Death in 507 Patients. Infection Control and Hospital Epidemiology. 2015; 36: 1251–1260.

[Google Scholar]

Rezkalla SH, Kloner RA. Influenza-related viral myocarditis. Wisconsin Medical Journal. 2010; 109: 209–213.

[Google Scholar]

Simonsen L, Fukuda K, Schonberger LB, Cox NJ. The impact of influenza epidemics on hospitalizations. The Journal of Infectious Diseases. 2000; 181: 831–837.

[Google Scholar]

Park M, Wu P, Goldstein E, Joo Kim W, Cowling BJ. Influenza-Associated Excess Mortality in South Korea. American Journal of Preventive Medicine. 2016; 50: e111–e119.

[Google Scholar]

Zheng J, Huo X, Huai Y, Xiao L, Jiang H, Klena J, et al. Epidemiology, Seasonality and Treatment of Hospitalized Adults and Adolescents with Influenza in Jingzhou, China, 2010-2012. PLoS ONE. 2016; 11: e0150713.

[Google Scholar]

Kalil AC, Thomas PG. Influenza virus-related critical illness: pathophysiology and epidemiology. Critical Care. 2019; 23: 258.

[Google Scholar]

Hsieh Y, Tsao K, Huang C, Chang K, Huang Y, Gong Y. Clinical characteristics of patients with laboratory-confirmed influenza a(H1N1)pdm09 during the 2013/2014 and 2015/2016 clade 6B/6B.1/6B.2-predominant outbreaks. Scientific Reports. 2018; 8: 15636.

[Google Scholar]

Yin R, Luusua E, Dabrowski J, Zhang Y, Kwoh CK. Tempel: time-series mutation prediction of influenza a viruses via attention-based recurrent neural networks. Bioinformatics. 2020; 36: 2697–2704.

[Google Scholar]

Yin R, Zhou X, Rashid S, Kwoh CK. HopPER: an adaptive model for probability estimation of influenza reassortment through host prediction. BMC Medical Genomics. 2020; 13: 9.

[Google Scholar]

Shlomai A, Nutman A, Kotlovsky T, Schechner V, Carmeli Y, Guzner-Gur H. Predictors of pandemic (H1N1) 2009 virus positivity and adverse outcomes among hospitalized patients with a compatible syndrome. The Israel Medical Association Journal. 2010; 12: 622–627.

[Google Scholar]

Cunha BA, Syed U, Mickail N, Strollo S. Rapid clinical diagnosis in fatal swine influenza (H1N1) pneumonia in an adult with negative rapid influenza diagnostic tests (RIDTs): diagnostic swine influenza triad. Heart & Lung. 2010; 39: 78–86.

[Google Scholar]

Teng F, Wan T, Guo S, Liu X, Cai J, Qi X, et al. Outcome prediction using the Mortality in Emergency Department Sepsis score combined with procalcitonin for influenza patients. Medicina ClíNica. 2019; 153: 411–417.

[Google Scholar]

Ho Y, Wang J, Wang J, Wu U, Chang C, Wu H, et al. Prognostic factors for fatal adult influenza pneumonia. The Journal of Infection. 2009; 58: 439–445.

[Google Scholar]

Morton B, Nweze K, O’Connor J, Turton P, Joekes E, Blakey JD, et al. Oxygen exchange and C-reactive protein predict safe discharge in patients with H1N1 influenza. QJM: Monthly Journal of the Association of Physicians. 2017; 110: 227–232.

[Google Scholar]

Cho WH, Kim YS, Jeon DS, Kim JE, Kim KI, Seol HY, et al. Outcome of pandemic H1N1 pneumonia: clinical and radiological findings for severity assessment. The Korean Journal of Internal Medicine. 2011; 26: 160–167.

[Google Scholar]

Khan Z, Hulme J, Sherwood N. An assessment of the validity of SOFA score based triage in H1N1 critically ill patients during an influenza pandemic. Anaesthesia. 2009; 64: 1283–1288.

[Google Scholar]

Chung J, Hsu C, Chen J, Chen W, Lin H, Guo H, et al. Geriatric influenza death (GID) score: a new tool for predicting mortality in older people with influenza in the emergency department. Scientific Reports. 2018; 8: 9312.

[Google Scholar]

Adeniji KA, Cusack R. The Simple Triage Scoring System (STSS) successfully predicts mortality and critical care resource utilization in H1N1 pandemic flu: a retrospective analysis. Critical Care. 2011; 15: R39.

[Google Scholar]

Chung J, Hsu C, Chen J, Chen W, Lin H, Guo H, et al. Shock index predicted mortality in geriatric patients with influenza in the emergency department. The American Journal of Emergency Medicine. 2019; 37: 391–394.

[Google Scholar]

Oh WS, Lee S, Lee C, Hur J, Hur A, Park YS, et al. A prediction rule to identify severe cases among adult patients hospitalized with pandemic influenza a (H1N1) 2009. Journal of Korean Medical Science. 2011; 26: 499–506.

[Google Scholar]

Yin R, Tran VH, Zhou X, Zheng J, Kwoh CK. Predicting antigenic variants of H1N1 influenza virus based on epidemics and pandemics using a stacking model. PLoS ONE. 2018; 13: e0207777.

[Google Scholar]

Dreyer NA, Bryant A, Velentgas P. The GRACE Checklist: a Validated Assessment Tool for High Quality Observational Studies of Comparative Effectiveness. Journal of Managed Care & Specialty Pharmacy. 2016; 22: 1107–1113.

[Google Scholar]

Wong CM, Yang L, Chan KP, Chan WM, Song L, Lai HK, et al. Cigarette smoking as a risk factor for influenza-associated mortality: evidence from an elderly cohort. Influenza and other Respiratory Viruses. 2013; 7: 531–539.

[Google Scholar]

Louie JK, Acosta M, Samuel MC, Schechter R, Vugia DJ, Harriman K, et al. A novel risk factor for a novel virus: obesity and 2009 pandemic influenza a (H1N1). Clinical Infectious Diseases. 2011; 52: 301–312.

[Google Scholar]

Lopez-Delgado JC, Rovira A, Esteve F, Rico N, Mañez Mendiluce R, Ballús Noguera J, et al. Thrombocytopenia as a mortality risk factor in acute respiratory failure in H1N1 influenza. Swiss Medical Weekly. 2013; 143: w13788.

[Google Scholar]

Demirjian SG, Raina R, Bhimraj A, Navaneethan SD, Gordon SM, Schreiber MJ, et al. 2009 influenza a infection and acute kidney injury: incidence, risk factors, and complications. American Journal of Nephrology. 2011; 34: 1–8.

[Google Scholar]

Bijani B, Pahlevan AA, Qasemi-Barqi R, Jahanihashemi H. Metabolic syndrome as an independent risk factor of hypoxaemia in influenza A (H1N1) 2009 pandemic. Le Infezioni in Medicina. 2016; 24: 123–130.

[Google Scholar]

Atamna A, Daskal R, Babich T, Ayada G, Ben-Zvi H, Elis A, et al. The impact of obesity on seasonal influenza: a single-center, retrospective study conducted in Israel. European Journal of Clinical Microbiology & Infectious Diseases. 2021; 40: 1471–1476.

[Google Scholar]

Fujikura Y, Kawano S, Kouzaki Y, Shinoda M, Hara Y, Shinkai M, et al. Mortality and severity evaluation by routine pneumonia prediction models among Japanese patients with 2009 pandemic influenza a (H1N1) pneumonia. Respiratory Investigation. 2014; 52: 280–287.

[Google Scholar]

Capelastegui A, Quintana JM, Bilbao A, España PP, Garin O, Alonso J, et al. Score to identify the severity of adult patients with influenza a (H1N1) 2009 virus infection at hospital admission. European Journal of Clinical Microbiology & Infectious Diseases. 2012; 31: 2693–2701.

[Google Scholar]

Choi WI, Yim JJ, Park J, Kim SC, Na MJ, Lee WY, et al. Clinical characteristics and outcomes of H1N1-associated pneumonia among adults in South Korea. International Journal of Tuberculosis and Lung Disease. 2011; 15: 270–275, i.

[Google Scholar]

Cheung W, Myburgh J, Seppelt IM, Parr MJ, Blackwell N, Demonte S, et al. Development and evaluation of an influenza pandemic intensive care unit triage protocol. Critical Care and Resuscitation. 2012; 14: 185–190.

[Google Scholar]

Rodriguez-Noriega E, Gonzalez-Diaz E, Morfin-Otero R, Gomez-Abundis GF, Briseño-Ramirez J, Perez-Gomez HR, et al. Hospital triage system for adult patients using an influenza-like illness scoring system during the 2009 pandemic–Mexico. PLoS ONE. 2010; 5: e10658.

[Google Scholar]

Zhang P, Cao B, Li X, Liang L, Yang S, Gu L, et al. Risk factors for adult death due to 2009 pandemic influenza a (H1N1) virus infection: a 2151 severe and critical cases analysis. Chinese Medical Journal. 2013; 126: 2222–2228.

[Google Scholar]

Riquelme R, Jiménez P, Videla AJ, Lopez H, Chalmers J, Singanayagam A, et al. Predicting mortality in hospitalized patients with 2009 H1N1 influenza pneumonia. The International Journal of Tuberculosis and Lung Disease. 2011; 15: 542–546.

[Google Scholar]

Pereira JM, Moreno RP, Matos R, Rhodes A, Martin-Loeches I, Cecconi M, et al. Severity assessment tools in ICU patients with 2009 influenza a (H1N1) pneumonia. Clinical Microbiology and Infection. 2012; 18: 1040–1048.

[Google Scholar]

Commons RJ, Denholm J. Triaging pandemic flu: pneumonia severity scores are not the answer. The International Journal of Tuberculosis and Lung Disease. 2012; 16: 670–673.

[Google Scholar]

Kiliç H, Kanbay A, Karalezli A, Hasanoğlu HC, Ateş C. Clinical characteristics of 75 pandemic H1N1 influenza patientsfrom Turkey; risk factors for fatality. Turkish Journal of Medical Sciences. 2015; 45: 562–567.

[Google Scholar]

Brandão-Neto RA, Goulart AC, Santana ANC, Martins HS, Ribeiro SCC, Ho LY, et al. The role of pneumonia scores in the emergency room in patients infected by 2009 H1N1 infection. European Journal of Emergency Medicine. 2012; 19: 200–202.

[Google Scholar]

Muller MP, McGeer AJ, Hassan K, Marshall J, Christian M. Evaluation of pneumonia severity and acute physiology scores to predict ICU admission and mortality in patients hospitalized for influenza. PLoS ONE. 2010; 5: e9563.

[Google Scholar]

Challen K, Bright J, Bentley A, Walter D. Physiological-social score (PMEWS) vs. CURB-65 to triage pandemic influenza: a comparative validation study using community-acquired pneumonia as a proxy. BMC Health Services Research. 2007; 7: 33.

[Google Scholar]

Rowan KM, Harrison DA, Walsh TS, McAuley DF, Perkins GD, Taylor BL, et al. The Swine Flu Triage (SwiFT) study: development and ongoing refinement of a triage tool to provide regular information to guide immediate policy and practice for the use of critical care services during the H1N1 swine influenza pandemic. Health Technol Assess. 2010; 14: 335–492.

[Google Scholar]

Bjarnason A, Thorleifsdottir G, Löve A, Gudnason JF, Asgeirsson H, Hallgrimsson KL, et al. Severity of influenza a 2009 (H1N1) pneumonia is underestimated by routine prediction rules. Results from a prospective, population-based study. PLoS ONE. 2012; 7: e46816.

[Google Scholar]

Shi SJ, Li H, Liu M, Liu YM, Zhou F, Liu B, et al. Mortality prediction to hospitalized patients with influenza pneumonia: PO2 /FiO2 combined lymphocyte count is the answer. The Clinical Respiratory Journal. 2017; 11: 352–360.

[Google Scholar]

Pawelka E, Karolyi M, Daller S, Kaczmarek C, Laferl H, Niculescu I, et al. Influenza virus infection: an approach to identify predictors for in-hospital and 90-day mortality from patients in Vienna during the season 2017/18. Infection. 2020; 48: 51–56.

[Google Scholar]

Franchini M, Veneri D, Lippi G. Thrombocytopenia and infections. Expert Review of Hematology. 2017; 10: 99–106.

[Google Scholar]

Siontis GCM, Tzoulaki I, Castaldi PJ, Ioannidis JPA. External validation of new risk prediction models is infrequent and reveals worse prognostic discrimination. Journal of Clinical Epidemiology. 2015; 68: 25–34.

[Google Scholar]

Debray TPA, Riley RD, Rovers MM, Reitsma JB, Moons KGM. Individual participant data (IPD) meta-analyses of diagnostic and prognostic modeling studies: guidance on their use. PLoS Medicine. 2015; 12: e1001886.

[Google Scholar]

Steyerberg EW, Harrell FE. Prediction models need appropriate internal, internal-external, and external validation. Journal of Clinical Epidemiology. 2016; 69: 245–247.

[Google Scholar]

Riley RD, Ensor J, Snell KIE, Debray TPA, Altman DG, Moons KGM, et al. External validation of clinical prediction models using big datasets from e-health records or IPD meta-analysis: opportunities and challenges. British Medical Journal. 2016; 353: i3140.

[Google Scholar]

Collins GS, Moons KGM. Comparing risk prediction models. British Medical Journal. 2012; 344: e3186.

[Google Scholar]

Wynants L, Kent DM, Timmerman D, Lundquist CM, Van Calster B. Untapped potential of multicenter studies: a review of cardiovascular risk prediction models revealed inappropriate analyses and wide variation in reporting. Diagnostic and Prognostic Research. 2019; 3: 6.

[Google Scholar]

Wynants L, Riley RD, Timmerman D, Van Calster B. Random-effects meta-analysis of the clinical utility of tests and prediction models. Statistics in Medicine. 2019; 37: 2034–2052.

[Google Scholar]

Riley RD, Ensor J, Snell KIE, Harrell FE Jr, Martin GP, Reitsma JB, et al. Calculating the sample size required for developing a clinical prediction model. British Medical Journal. 2020; 368: m441.

[Google Scholar]

Guo L, Wei D, Zhang X, Wu Y, Li Q, Zhou M, et al. Clinical Features Predicting Mortality Risk in Patients With Viral Pneumonia: The MuLBSTA Score. Frontiers in Microbiology. 2019; 10: 2752.

[Google Scholar]

Kuba K, Imai Y, Rao S, Gao H, Guo F, Guan B, et al. A crucial role of angiotensin converting enzyme 2 (ACE2) in SARS coronavirus-induced lung injury. Nature Medicine. 2005; 11: 875–879.

[Google Scholar]

Xu Z, Shi L, Wang Y, Zhang J, Huang L, Zhang C, et al. Pathological findings of COVID-19 associated with acute respiratory distress syndrome. The Lancet Respiratory Medicine. 2020; 8: 420–422.

[Google Scholar]

Damen JAAG, Hooft L, Schuit E, Debray TPA, Collins GS, Tzoulaki I, et al. Prediction models for cardiovascular disease risk in the general population: systematic review. British Medical Journal. 2016; 353: i2416.

[Google Scholar]