Clinical and laboratory predictors of influenza infection among individuals with influenza-like illness presenting to an urban Thai hospital over a five-year period

Do all Emergency Room Patients With Influenza-like Symptoms Need Blood Cultures? A Retrospective Cohort Study of 2 Annual Influenza Seasons

In this retrospective cohort study, we evaluated risk factors for bacteremia in emergency department patients presenting with influenza-like symptoms during influenza epidemic seasons. In patients without fever, chronic heart or chronic liver disease, blood culture collection might be omitted.

Novel graph-based machine-learning technique for viral infectious diseases: application to influenza and hepatitis diseases

BACKGROUND

Most infectious diseases are caused by viruses, fungi, bacteria and parasites. Their ability to easily infect humans and trigger large-scale epidemics makes them a public health concern. Methods for early detection of these diseases have been developed; however, they are hindered by the absence of a unified, interoperable and reusable model. This study seeks to create a holistic and real-time model for swift, preliminary detection of infectious diseases using symptoms and additional clinical data.

MATERIALS AND METHODS

In this study, we present a medical knowledge graph (MKG) that leverages multiple data sources to analyse connections between different nodes. Medical ontologies were used to enhance the MKG. We applied various graph algorithms to extract key features. The performance of multiple machine-learning (ML) techniques for influenza and hepatitis detection was assessed, selecting multi-layer perceptron (MLP) and random forest (RF) models due to their superior outcomes. The hyperparameters of both graph-based ML models were automatically fine-tuned.

RESULTS

Both the graph-based MLP and RF models showcased the least loss and error rates, along with the most specific, accurate recall, precision and F1 scores. Their Matthews correlation coefficients were also optimal. When compared with existing ML techniques and findings from the literature, these graph-based ML models manifested superior detection accuracy.

CONCLUSIONS

The graph-based MLP and RF models effectively diagnosed influenza and hepatitis, respectively. This underlines the potential of graph data science in enhancing ML model performance and uncovering concealed relationships in the MKG.

Machine-learning-algorithms-based diagnostic model for influenza A in children

BACKGROUND

At present, nucleic acid testing is the gold standard for diagnosing influenza A, however, this method is expensive, time-consuming, and unsuitable for promotion and use in grassroots hospitals. This study aimed to establish a diagnostic model that could accurately, quickly, and simply distinguish between influenza A and influenza like diseases.

METHODS

Patients with influenza-like symptoms were recruited between December 2019 and August 2023 at the Children's Hospital Affiliated to Shandong University and basic information, nasopharyngeal swab and blood routine test data were included. Computer algorithms including random forest, GBDT, XGBoost and logistic regression (LR) were used to create the diagnostic model, and their performance was evaluated using the validation data sets.

RESULTS

A total of 4188 children with influenza-like symptoms were enrolled, of which 1992 were nucleic acid test positive and 2196 were matched negative. The diagnostic models based on the random forest, GBDT, XGBoost and logistic regression algorithms had AUC values of 0.835,0.872,0.867 and 0.784, respectively. The top 5 important features were lymphocyte (LYM) count, age, serum amyloid A (SAA), white blood cells (WBC) count and platelet-to-lymphocyte ratio (PLR). GBDT model had the best performance, the sensitivity and specificity were 77.23% and 80.29%, respectively.

CONCLUSIONS

A computer algorithm diagnosis model of influenza A in children based on blood routine test data was established, which could identify children with influenza A more accurately in the early stage, and was easy to popularize.

Developing and validating clinical features-based machine learning algorithms to predict influenza infection in influenza-like illness patients

BACKGROUND

Seasonal influenza poses a significant risk, and patients can benefit from early diagnosis and treatment. However, underdiagnosis and undertreatment remain widespread. We developed and compared clinical feature-based machine learning (ML) algorithms that can accurately predict influenza infection in emergency departments (EDs) among patients with influenza-like illness (ILI).

MATERIAL AND METHODS

We conducted a prospective cohort study in five EDs in the US and Taiwan from 2015 to 2020. Adult patients visiting the EDs with symptoms of ILI were recruited and tested by real-time RT-PCR for influenza. We evaluated seven ML algorithms and compared their results with previously developed clinical prediction models.

RESULTS

Out of the 2189 enrolled patients, 1104 tested positive for influenza. The eXtreme Gradient Boosting achieved superior performance with an area under the receiver operating characteristic curve of 0.82 (95% confidence interval [CI] = 0.79-0.85), with a sensitivity of 0.92 (95% CI = 0.88-0.95), specificity of 0.89 (95% CI = 0.86-0.92), and accuracy of 0.72 (95% CI = 0.69-0.76) in the testing set over cut-offs of 0.4, 0.6 and 0.5, respectively. These results were superior to those of previously proposed clinical prediction models. The model interpretation revealed that body temperature, cough, rhinorrhea, and exposure history were positively associated with and the days of illness and influenza vaccine were negatively associated with influenza infection. We also found the week of the influenza season, pulse rate, and oxygen saturation to be associated with influenza infection.

CONCLUSIONS

The clinical feature-based ML model outperformed conventional models for predicting influenza infection.

Distinguishing SARS-CoV-2 Infection and Non-SARS-CoV-2 Viral Infections in Adult Patients through Clinical Score Tools

This study aimed to determine distinguishing predictors and develop a clinical score to differentiate COVID-19 and common viral infections (influenza, respiratory syncytial virus (RSV), dengue, chikungunya (CKV), and zika (ZKV)). This retrospective study enrolled 549 adults (100 COVID-19, 100 dengue, 100 influenza, 100 RSV, 100 CKV, and 49 ZKV) during the period 2017−2020. CKV and ZKV infections had specific clinical features (i.e., arthralgia and rash); therefore, these diseases were excluded. Multiple binary logistic regression models were fitted to identify significant predictors, and two scores were developed differentiating influenza/RSV from COVID-19 (Flu-RSV/COVID) and dengue from COVID-19 (Dengue/COVID). The five independent predictors of influenza/RSV were age > 50 years, the presence of underlying disease, rhinorrhea, productive sputum, and lymphocyte count < 1000 cell/mm3. Likewise, the five independent predictors of dengue were headache, myalgia, no cough, platelet count < 150,000/mm3, and lymphocyte count < 1000 cell/mm3. The Flu-RSV/COVID score (cut-off value of 4) demonstrated 88% sensitivity and specificity for predicting influenza/RSV (AUROC = 0.94). The Dengue/COVID score (cut-off value of 4) achieved 91% sensitivity and 94% specificity for differentiating dengue and COVID-19 (AUROC = 0.98). The Flu-RSV/COVID and Dengue/COVID scores had a high discriminative ability for differentiating influenza/RSV or dengue infection and COVID-19. The further validation of these scores is needed to ensure their utility in clinical practice.

Applying symptom dynamics to accurately predict influenza virus infection: An international multicenter influenza-like illness surveillance study

BACKGROUND

Public health organizations have recommended various definitions of influenza-like illnesses under the assumption that the symptoms do not change during influenza virus infection. To explore the relationship between symptoms and influenza over time, we analyzed a dataset from an international multicenter prospective emergency department (ED)-based influenza-like illness cohort study.

METHODS

We recruited patients in the US and Taiwan between 2015 and 2020 with: (1) flu-like symptoms (fever and cough, headache, or sore throat), (2) absence of any of the respiratory infection symptoms, or (3) positive laboratory test results for influenza from the current ED visit. We evaluated the association between the symptoms and influenza virus infection on different days of illness. The association was evaluated among different subgroups, including different study countries, influenza subtypes, and only patients with influenza.

RESULTS

Among the 2471 recruited patients, 45.7% tested positive for influenza virus. Cough was the most predictive symptom throughout the week (odds ratios [OR]: 7.08-11.15). In general, all symptoms were more predictive during the first 2 days (OR: 1.55-10.28). Upper respiratory symptoms, such as sore throat and productive cough, and general symptoms, such as body ache and fatigue, were more predictive in the first half of the week (OR: 1.51-3.25). Lower respiratory symptoms, such as shortness of breath and wheezing, were more predictive in the second half of the week (OR: 1.52-2.52). Similar trends were observed for most symptoms in the different subgroups.

CONCLUSIONS

The time course is an important factor to be considered when evaluating the symptoms of influenza virus infection.

Assessment of potential factors associated with the sensitivity and specificity of Sofia Influenza A+B Fluorescent Immunoassay in an ambulatory care setting

Background

Seasonal influenza leads to an increase in outpatient clinic visits. Timely, accurate, and affordable testing could facilitate improved treatment outcomes. Rapid influenza diagnostic tests (RIDTs) provide results in as little as 15 minutes and are relatively inexpensive, but have reduced sensitivity when compared to RT-PCR. The contributions of multiple factors related to test performance are not well defined for ambulatory care settings. We assessed clinical and laboratory factors that may affect the sensitivity and specificity of Sofia Influenza A+B Fluorescence Immunoassay.

Study design

We performed a post-hoc assessment of surveillance data amassed over seven years from five primary care clinics. We analyzed 4,475 paired RIDT and RT-PCR results from specimens collected from patients presenting with respiratory symptoms and examined eleven potential factors with additional sub-categories that could affect RIDT sensitivity.

Results

In an unadjusted analysis, greater sensitivity was associated with the presence of an influenza-like illness (ILI), no other virus detected, no seasonal influenza vaccination, younger age, lower cycle threshold value, fewer days since illness onset, nasal discharge, stuffy nose, and fever. After adjustment, presence of an ILI, younger age, fewer days from onset, no co-detection, and presence of a nasal discharge maintained significance.

Conclusion

Clinical and laboratory factors may affect RIDT sensitivity. Identifying potential factors during point-of-care testing could aid clinicians in appropriately interpreting negative influenza RIDT results.

How are reasons for encounter associated with influenza-like illness and acute respiratory infection diagnoses and interventions? A cohort study in eight Italian general practice populations

Background

Influenza-like illness (ILI) and Acute Respiratory Infections (ARI) are a considerable health problem in Europe. Most diagnoses are made by family physicians (FPs) and based on symptoms and clinical signs rather than on diagnostic testing. The International Classification of Primary Care (ICPC) advocates that FPs record patients’ ‘Reasons for Encounters’ (RfEs) as they are presented to them.

This study analyses the association of patients’ RfEs with FPs’ diagnoses of ILI and ARI diagnoses and FPs’ management of those patients.

Methods

Cohort study of practice populations. Over a 4-month period during the winter season 2013–14, eight FPs recorded ILI and ARI patients’ RfEs and how they were managed. FPs recorded details of their patients using the ICPC format, collecting data in an Episode of Care (EoC) structure.

Results

There were 688 patients diagnosed as having ILI; between them they presented with a total of 2,153 RfEs, most commonly fever (79.7%), cough (59.7%) and pain (33.0%).

The 848 patients with ARI presented with a total of 1,647 RfEs, most commonly cough (50.4%), throat symptoms (25.9%) and fever (19.9%). For patients with ILI, 37.0% of actions were related to medication for respiratory symptoms; this figure was 38.4% for patients with ARI. FPs referred six patients to specialists or hospitals (0.39% of all patients diagnosed with ILI and ARI).

Conclusions

In this study of patients with ILI and ARI, less than half received a prescription from their FPs, and the illnesses were mainly managed in primary care, with few patients’ needing referral. The ICPC classification allowed a standardised data collection system, providing documentary evidence of the management of those diseases.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12875-021-01519-4.

A brief review of influenza virus infection

Influenza is an acute viral respiratory infection that affects all age groups and is associated with high mortality during pandemics, epidemics, and sporadic outbreaks. Nearly 10% of the world's population is affected by influenza annually, with about half a million deaths each year. Influenza vaccination is the most effective method for preventing influenza infection and its complications. The influenza vaccine's efficacy varies each season based on the circulating influenza strains and vaccine uptake rates. Currently, three antiviral drugs targeting the influenza virus surface glycoprotein neuraminidase are available for treatment and prophylaxis of disease. Given the significant burden of influenza infection globally, this review is focused on the latest findings in the etiology, epidemiology, transmission, clinical manifestation, diagnosis, prevention, and treatment of influenza.

Performance of case definitions and clinical predictors for influenza surveillance among patients followed in a rural cohort in Senegal

Background

Influenza is a major cause of morbidity and mortality in Africa. However, a lack of epidemiological data remains for this pathology, and the performances of the influenza-like illness (ILI) case definitions used for sentinel surveillance have never been evaluated in Senegal. This study aimed to i) assess the performance of three different ILI case definitions, adopted by the WHO, USA-CDC (CDC) and European-CDC (ECDC) and ii) identify clinical factors associated with a positive diagnosis for Influenza in order to develop an algorithm fitted for the Senegalese context.

Methods

All 657 patients with a febrile pathological episode (FPE) between January 2013 and December 2016 were followed in a cohort study in two rural villages in Senegal, accounting for 1653 FPE observations with nasopharyngeal sampling and influenza virus screening by rRT-PCR. For each FPE, general characteristics and clinical signs presented by patients were collected. Sensitivity, Specificity, Positive Predictive Value (PPV) and Negative Predictive Value (NPV) for the three ILI case definitions were assessed using PCR result as the reference test. Associations between clinical signs and influenza infection were analyzed using logistic regression with generalized estimating equations. Sore throat, arthralgia or myalgia were missing for children under 5 years.

Results

WHO, CDC and ECDC case definitions had similar sensitivity (81.0%; 95%CI: 77.0–85.0) and NPV (91.0%; 95%CI: 89.0–93.1) while the WHO and CDC ILI case definitions had the highest specificity (52.0%; 95%CI: 49.1–54.5) and PPV (32.0%; 95%CI: 30.0–35.0). These performances varied by age groups. In children < 5 years, the significant predictors of influenza virus infection were cough and nasal discharge. In patients from 5 years, cough, nasal discharge, sore throat and asthenia grade 3 best predicted influenza infection. The addition of “nasal discharge” as a symptom to the WHO case definition decreased sensitivity but increased specificity, particularly in the pediatric population.

Conclusion

In summary, all three definitions studies (WHO, ECDC & CDC) have similar performance, even by age group. The revised WHO ILI definition could be chosen for surveillance purposes for its simplicity. Symptomatic predictors of influenza virus infection vary according the age group.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-020-05724-x.

Clinical signs predictive of influenza virus infection in Cameroon

Influenza virus accounts for majority of respiratory virus infections in Cameroon. According to the World Health Organization (WHO), influenza-like illnesses (ILI) are identified by a measured temperature of ≥38°C and cough, with onset within the past 10 days. Other symptoms could as well be observed however, none of these are specific to influenza alone. This study aimed to determine symptom based predictors of influenza virus infection in Cameroon. Individuals with ILI were recruited from 2009-2018 in sentinel sites of the influenza surveillance system in Cameroon according to the WHO case definition. Individual data collection forms accompanied each respiratory sample and contained clinical data. Samples were analyzed for influenza using the gold standard assay. Two statistical methods were compared to determine the most reliable clinical predictors of influenza virus activity in Cameroon: binomial logistic predictive model and random forest model. Analyses were performed in R version 3.5.2. A total of 11816 participants were recruited, of which, 24.0% were positive for influenza virus. Binomial logistic predictive model revealed that the presence of cough, rhinorrhoea, headache and myalgia are significant predictors of influenza positivity. The prediction model had a sensitivity of 75.6%, specificity of 46.6% and AUC of 66.7%. The random forest model categorized the reported symptoms according to their degree of importance in predicting influenza virus infection. Myalgia had a 2-fold higher value in predicting influenza virus infection compared to any other symptom followed by arthralgia, head ache, rhinorrhoea and sore throat. The model had a OOB error rate of 25.86%. Analysis showed that the random forest model had a better performance over the binomial regression model in predicting influenza infection. Rhinorrhoea, headache and myalgia were symptoms reported by both models as significant predictors of influenza infection in Cameroon. These symptoms could be used by clinicians in their decision to treat patients.

The Use of Test-negative Controls to Monitor Vaccine Effectiveness: A Systematic Review of Methodology

BACKGROUND

The test-negative design is an increasingly popular approach for estimating vaccine effectiveness (VE) due to its efficiency. This review aims to examine published test-negative design studies of VE and to explore similarities and differences in methodological choices for different diseases and vaccines.

METHODS

We conducted a systematic search on PubMed, Web of Science, and Medline, for studies reporting the effectiveness of any vaccines using a test-negative design. We screened titles and abstracts and reviewed full texts to identify relevant articles. We created a standardized form for each included article to extract information on the pathogen of interest, vaccine(s) being evaluated, study setting, clinical case definition, choices of cases and controls, and statistical approaches used to estimate VE.

RESULTS

We identified a total of 348 articles, including studies on VE against influenza virus (n = 253), rotavirus (n = 48), pneumococcus (n = 24), and nine other pathogens. Clinical case definitions used to enroll patients were similar by pathogens of interest but the sets of symptoms that defined them varied substantially. Controls could be those testing negative for the pathogen of interest, those testing positive for nonvaccine type of the pathogen of interest, or a subset of those testing positive for alternative pathogens. Most studies controlled for age, calendar time, and comorbidities.

CONCLUSIONS

Our review highlights similarities and differences in the application of the test-negative design that deserve further examination. If vaccination reduces disease severity in breakthrough infections, particular care must be taken in interpreting vaccine effectiveness estimates from test-negative design studies.

Development and validation of clinical prediction models to distinguish influenza from other viruses causing acute respiratory infections in children and adults

Predictive models have been developed for influenza but have seldom been validated. Typically they have focused on patients meeting a definition of infection that includes fever. Less is known about how models perform when more symptoms are considered. We, therefore, aimed to create and internally validate predictive scores of acute respiratory infection (ARI) symptoms to diagnose influenza virus infection as confirmed by polymerase chain reaction (PCR) from respiratory specimens. Data from a completed trial to study the indirect effect of influenza immunization in Hutterite communities were randomly split into two independent groups for model derivation and validation. We applied different multivariable modelling techniques and constructed Receiver Operating Characteristics (ROC) curves to determine predictive indexes at different cut-points. From 2008–2011, 3288 first seasonal ARI episodes and 321 (9.8%) influenza positive events occurred in 2202 individuals. In children up to 17 years, the significant predictors of influenza virus infection were fever, chills, and cough along with being of age 6 years and older. In adults, presence of chills and cough but not fever were highly specific for influenza virus infection (sensitivity 30%, specificity 96%). Performance of the models in the validation set was not significantly different. The predictors were consistently found to be significant irrespective of the multivariable technique. Symptomatic predictors of influenza virus infection vary between children and adults. The scores could assist clinicians in their test and treat decisions but the results need to be externally validated prior to application in clinical practice.