Forecasting severe respiratory disease hospitalizations using machine learning algorithms.

BACKGROUND: Forecasting models predicting trends in hospitalization rates have the potential to inform hospital management during seasonal epidemics of respiratory diseases and the associated surges caused by acute hospital admissions. Hospital bed requirements for elective surgery could be better planned if it were possible to foresee upcoming peaks in severe respiratory illness admissions. Forecasting models can also guide the use of intervention strategies to decrease the spread of respiratory pathogens and thus prevent local health system overload. In this study, we explore the capability of forecasting models to predict the number of hospital admissions in Auckland, New Zealand, within a three-week time horizon. Furthermore, we evaluate probabilistic forecasts and the impact on model performance when integrating laboratory data describing the circulation of respiratory viruses. METHODS: The dataset used for this exploration results from active hospital surveillance, in which the World Health Organization Severe Acute Respiratory Infection (SARI) case definition was consistently used. This research nurse-led surveillance has been implemented in two public hospitals in Auckland and provides a systematic laboratory testing of SARI patients for nine respiratory viruses, including influenza, respiratory syncytial virus, and rhinovirus. The forecasting strategies used comprise automatic machine learning, one of the most recent generative pre-trained transformers, and established artificial neural network algorithms capable of univariate and multivariate forecasting. RESULTS: We found that machine learning models compute more accurate forecasts in comparison to naïve seasonal models. Furthermore, we analyzed the impact of reducing the temporal resolution of forecasts, which decreased the model error of point forecasts and made probabilistic forecasting more reliable. An additional analysis that used the laboratory data revealed strong season-to-season variations in the incidence of respiratory viruses and how this correlates with total hospitalization cases. These variations could explain why it was not possible to improve forecasts by integrating this data. CONCLUSIONS: Active SARI surveillance and consistent data collection over time enable these data to be used to predict hospital bed utilization. These findings show the potential of machine learning as support for informing systems for proactive hospital management.

Effect of Early High-Flow Nasal Oxygen vs Standard Oxygen Therapy on Length of Hospital Stay in Hospitalized Children With Acute Hypoxemic Respiratory Failure: The PARIS-2 Randomized Clinical Trial.

Importance: Nasal high-flow oxygen therapy in infants with bronchiolitis and hypoxia has been shown to reduce the requirement to escalate care. The efficacy of high-flow oxygen therapy in children aged 1 to 4 years with acute hypoxemic respiratory failure without bronchiolitis is unknown. Objective: To determine the effect of early high-flow oxygen therapy vs standard oxygen therapy in children with acute hypoxemic respiratory failure. Design, Setting, and Participants: A multicenter, randomized clinical trial was conducted at 14 metropolitan and tertiary hospitals in Australia and New Zealand, including 1567 children aged 1 to 4 years (randomized between December 18, 2017, and March 18, 2020) requiring hospital admission for acute hypoxemic respiratory failure. The last participant follow-up was completed on March 22, 2020. Interventions: Enrolled children were randomly allocated 1:1 to high-flow oxygen therapy (n = 753) or standard oxygen therapy (n = 764). The type of oxygen therapy could not be masked, but the investigators remained blinded until the outcome data were locked. Main Outcomes and Measures: The primary outcome was length of hospital stay with the hypothesis that high-flow oxygen therapy reduces length of stay. There were 9 secondary outcomes, including length of oxygen therapy and admission to the intensive care unit. Children were analyzed according to their randomization group. Results: Of the 1567 children who were randomized, 1517 (97%) were included in the primary analysis (median age, 1.9 years [IQR, 1.4-3.0 years]; 732 [46.7%] were female) and all children completed the trial. The length of hospital stay was significantly longer in the high-flow oxygen group with a median of 1.77 days (IQR, 1.03-2.80 days) vs 1.50 days (IQR, 0.85-2.44 days) in the standard oxygen group (adjusted hazard ratio, 0.83 [95% CI, 0.75-0.92]; P < .001). Of the 9 prespecified secondary outcomes, 4 showed no significant difference. The median length of oxygen therapy was 1.07 days (IQR, 0.50-2.06 days) in the high-flow oxygen group vs 0.75 days (IQR, 0.35-1.61 days) in the standard oxygen therapy group (adjusted hazard ratio, 0.78 [95% CI, 0.70-0.86]). In the high-flow oxygen group, there were 94 admissions (12.5%) to the intensive care unit compared with 53 admissions (6.9%) in the standard oxygen group (adjusted odds ratio, 1.93 [95% CI, 1.35-2.75]). There was only 1 death and it occurred in the high-flow oxygen group. Conclusions and Relevance: Nasal high-flow oxygen used as the initial primary therapy in children aged 1 to 4 years with acute hypoxemic respiratory failure did not significantly reduce the length of hospital stay compared with standard oxygen therapy. Trial Registration: anzctr.org.au Identifier: ACTRN12618000210279.

COVID-19 and Infant Hospitalizations for Seasonal Respiratory Virus Infections, New Zealand, 2020.

In March 2020, a national elimination strategy for coronavirus disease was introduced in New Zealand. Since then, hospitalizations for lower respiratory tract infection among infants <2 years of age and cases of respiratory syncytial or influenza virus infection have dramatically decreased. These findings indicate additional benefits of coronavirus disease control strategies.

Prospective community programme versus parent-driven care to prevent respiratory morbidity in children following hospitalisation with severe bronchiolitis or pneumonia.

BACKGROUND: Hospitalisation with severe lower respiratory tract infection (LRTI) in early childhood is associated with ongoing respiratory symptoms and possible later development of bronchiectasis. We aimed to reduce this intermediate respiratory morbidity with a community intervention programme at time of discharge. METHODS: This randomised, controlled, single-blind trial enrolled children aged <2 years hospitalised for severe LRTI to 'intervention' or 'control'. Intervention was three monthly community clinics treating wet cough with prolonged antibiotics referring non-responders. All other health issues were addressed, and health resilience behaviours were encouraged, with referrals for housing or smoking concerns. Controls followed the usual pathway of parent-initiated healthcare access. After 24 months, all children were assessed by a paediatrician blinded to randomisation for primary outcomes of wet cough, abnormal examination (crackles or clubbing) or chest X-ray Brasfield score ≤22. FINDINGS: 400 children (203 intervention, 197 control) were enrolled in 2011-2012; mean age 6.9 months, 230 boys, 87% Maori/Pasifika ethnicity and 83% from the most deprived quintile. Final assessment of 321/400 (80.3%) showed no differences in presence of wet cough (33.9% intervention, 36.5% controls, relative risk (RR) 0.93, 95% CI 0.69 to 1.25), abnormal examination (21.7% intervention, 23.9% controls, RR 0.92, 95% CI 0.61 to 1.38) or Brasfield score ≤22 (32.4% intervention, 37.9% control, RR 0.85, 95% CI 0.63 to 1.17). Twelve (all intervention) were diagnosed with bronchiectasis within this timeframe. INTERPRETATION: We have identified children at high risk of ongoing respiratory disease following hospital admission with severe LRTI in whom this intervention programme did not change outcomes over 2 years. TRIAL REGISTRATION NUMBER: ACTRN12610001095055.

Risk Factors and Attack Rates of Seasonal Influenza Infection: Results of the Southern Hemisphere Influenza and Vaccine Effectiveness Research and Surveillance (SHIVERS) Seroepidemiologic Cohort Study.

Background: Understanding the attack rate of influenza infection and the proportion who become ill by risk group is key to implementing prevention measures. While population-based studies of antihemagglutinin antibody responses have been described previously, studies examining both antihemagglutinin and antineuraminidase antibodies are lacking. Methods: In 2015, we conducted a seroepidemiologic cohort study of individuals randomly selected from a population in New Zealand. We tested paired sera for hemagglutination inhibition (HAI) or neuraminidase inhibition (NAI) titers for seroconversion. We followed participants weekly and performed influenza polymerase chain reaction (PCR) for those reporting influenza-like illness (ILI). Results: Influenza infection (either HAI or NAI seroconversion) was found in 321 (35% [95% confidence interval, 32%-38%]) of 911 unvaccinated participants, of whom 100 (31%) seroconverted to NAI alone. Young children and Pacific peoples experienced the highest influenza infection attack rates, but overall only a quarter of all infected reported influenza PCR-confirmed ILI, and one-quarter of these sought medical attention. Seroconversion to NAI alone was higher among children aged <5 years vs those aged ≥5 years (14% vs 4%; P < .001) and among those with influenza B vs A(H3N2) virus infections (7% vs 0.3%; P < .001). Conclusions: Measurement of antineuraminidase antibodies in addition to antihemagglutinin antibodies may be important in capturing the true influenza infection rates.

Impact of the COVID-19 related border restrictions on influenza and other common respiratory viral infections in New Zealand.

BACKGROUND: New Zealand's (NZ) complete absence of community transmission of influenza and respiratory syncytial virus (RSV) after May 2020, likely due to COVID-19 elimination measures, provided a rare opportunity to assess the impact of border restrictions on common respiratory viral infections over the ensuing 2 years. METHODS: We collected the data from multiple surveillance systems, including hospital-based severe acute respiratory infection surveillance, SHIVERS-II, -III and -IV community cohorts for acute respiratory infection (ARI) surveillance, HealthStat sentinel general practice (GP) based influenza-like illness surveillance and SHIVERS-V sentinel GP-based ARI surveillance, SHIVERS-V traveller ARI surveillance and laboratory-based surveillance. We described the data on influenza, RSV and other respiratory viral infections in NZ before, during and after various stages of the COVID related border restrictions. RESULTS: We observed that border closure to most people, and mandatory government-managed isolation and quarantine on arrival for those allowed to enter, appeared to be effective in keeping influenza and RSV infections out of the NZ community. Border restrictions did not affect community transmission of other respiratory viruses such as rhinovirus and parainfluenza virus type-1. Partial border relaxations through quarantine-free travel with Australia and other countries were quickly followed by importation of RSV in 2021 and influenza in 2022. CONCLUSION: Our findings inform future pandemic preparedness and strategies to model and manage the impact of influenza and other respiratory viral threats.

Impact of the COVID-19 nonpharmaceutical interventions on influenza and other respiratory viral infections in New Zealand.

Stringent nonpharmaceutical interventions (NPIs) such as lockdowns and border closures are not currently recommended for pandemic influenza control. New Zealand used these NPIs to eliminate coronavirus disease 2019 during its first wave. Using multiple surveillance systems, we observed a parallel and unprecedented reduction of influenza and other respiratory viral infections in 2020. This finding supports the use of these NPIs for controlling pandemic influenza and other severe respiratory viral threats.

Impact of the COVID-19 nonpharmaceutical interventions on influenza and other respiratory viral infections in New Zealand.

Stringent nonpharmaceutical interventions (NPIs) such as lockdowns and border closures are not currently recommended for pandemic influenza control. New Zealand used these NPIs to eliminate coronavirus disease 2019 during its first wave. Using multiple surveillance systems, we observed a parallel and unprecedented reduction of influenza and other respiratory viral infections in 2020. This finding supports the use of these NPIs for controlling pandemic influenza and other severe respiratory viral threats.