Search

Abstract Background Accurately predicting hospital discharge events could help improve patient flow and the efficiency of healthcare delivery. However, using machine learning and diverse electronic health record (EHR) data for this task remains incompletely explored. Methods We used EHR data from February-2017 to January-2020 from Oxfordshire, UK to predict hospital discharges in the next 24 h. We fitted separate extreme gradient boosting models for elective and emergency admissions, trained on the first two years of data and tested on the final year of data. We examined individual-level and hospital-level model performance and evaluated the impact of training data size and recency, prediction time, and performance in subgroups. Results Our models achieve AUROCs of 0.87 and 0.86, AUPRCs of 0.66 and 0.64, and F1 scores of 0.61 and 0.59 for elective and emergency admissions, respectively. These models outperform a logistic regression model using the same features and are substantially better than a baseline logistic regression model with more limited features. Notably, the relative performance increase from adding additional features is greater than the increase from using a sophisticated model. Aggregating individual probabilities, daily total discharge estimates are accurate with mean absolute errors of 8.9% (elective) and 4.9% (emergency). The most informative predictors include antibiotic prescriptions, medications, and hospital capacity factors. Performance remains robust across patient subgroups and different training strategies, but is lower in patients with longer admissions and those who died in hospital. Conclusions Our findings highlight the potential of machine learning in optimising hospital patient flow and facilitating patient care and recovery.

Abstract Background Whole blood host transcript signatures show great potential for diagnosis of infectious and inflammatory illness, with most published signatures performing binary classification tasks. Barriers to clinical implementation include validation studies, and development of strategies that enable simultaneous, multiclass diagnosis of febrile illness based on gene expression. Methods We validated five distinct diagnostic signatures for paediatric infectious diseases in parallel using a single NanoString nCounter® experiment. We included a novel 3-transcript signature for childhood tuberculosis, and four published signatures which differentiate bacterial infection, viral infection, or Kawasaki disease from other febrile illnesses. Signature performance was assessed using receiver operating characteristic curve statistics. We also explored conceptual frameworks for multiclass diagnostic signatures, including additional transcripts found to be significantly differentially expressed in previous studies. Relaxed, regularised logistic regression models were used to derive two novel multiclass signatures: a mixed One-vs-All model (MOVA), running multiple binomial models in parallel, and a full-multiclass model. In-sample performance of these models was compared using radar-plots and confusion matrix statistics. Results Samples from 91 children were included in the study: 23 bacterial infections (DB), 20 viral infections (DV), 14 Kawasaki disease (KD), 18 tuberculosis disease (TB), and 16 healthy controls. The five signatures tested demonstrated cross-platform performance similar to their primary discovery-validation cohorts. The signatures could differentiate: KD from other diseases with area under ROC curve (AUC) of 0.897 [95% confidence interval: 0.822–0.972]; DB from DV with AUC of 0.825 [0.691–0.959] (signature-1) and 0.867 [0.753–0.982] (signature-2); TB from other diseases with AUC of 0.882 [0.787–0.977] (novel signature); TB from healthy children with AUC of 0.910 [0.808–1.000]. Application of signatures outside of their designed context reduced performance. In-sample error rates for the multiclass models were 13.3% for the MOVA model and 0.0% for the full-multiclass model. The MOVA model misclassified DB cases most frequently (18.7%) and TB cases least (2.7%). Conclusions Our study demonstrates the feasibility of NanoString technology for cross-platform validation of multiple transcriptomic signatures in parallel. This external cohort validated performance of all five signatures, including a novel sparse TB signature. Two exploratory multi-class models showed high potential accuracy across four distinct diagnostic groups.

Abstract Omicron variants of SARS-CoV-2 are globally dominant and infection rates are very high in children. We measure immune responses following Omicron BA.1/2 infection in children aged 6-14 years and relate this to prior and subsequent SARS-CoV-2 infection or vaccination. Primary Omicron infection elicits a weak antibody response with poor functional neutralizing antibodies. Subsequent Omicron reinfection or COVID-19 vaccination elicits increased antibody titres with broad neutralisation of Omicron subvariants. Prior pre-Omicron SARS-CoV-2 virus infection or vaccination primes for robust antibody responses following Omicron infection but these remain primarily focussed against ancestral variants. Primary Omicron infection thus elicits a weak antibody response in children which is boosted after reinfection or vaccination. Cellular responses are robust and broadly equivalent in all groups, providing protection against severe disease irrespective of SARS-CoV-2 variant. Immunological imprinting is likely to act as an important determinant of long-term humoral immunity, the future clinical importance of which is unknown.

Children and adolescents generally experience mild COVID-19. However, those with underlying physical health conditions are at a significantly increased risk of severe disease. Here, we present a comprehensive analysis of antibody and cellular responses in adolescents with severe neuro-disabilities who received COVID-19 vaccination with either ChAdOx1 (n=6) or an mRNA vaccine (mRNA-1273, n=8, BNT162b2, n=1). Strong immune responses were observed after vaccination and antibody levels and neutralisation titres were both higher after two doses. Both measures were also higher after mRNA vaccination and were further enhanced by prior natural infection where one vaccine dose was sufficient to generate peak antibody response. Robust T-cell responses were generated after dual vaccination and were also higher following mRNA vaccination. Early T-cells were characterised by a dominant effector-memory CD4+ T-cell population with a type-1 cytokine signature with additional production of IL-10. Antibody levels were well-maintained for at least 3 months after vaccination and 3 of 4 donors showed measurable neutralisation titres against the Omicron variant. T-cell responses also remained robust, with generation of a central/stem cell memory pool and showed strong reactivity against Omicron spike. These data demonstrate that COVID-19 vaccines display strong immunogenicity in adolescents and that dual vaccination, or single vaccination following prior infection, generate higher immune responses than seen after natural infection and develop activity against Omicron. Initial evidence suggests that mRNA vaccination elicits stronger immune responses than adenoviral delivery, although the latter is also higher than seen in adult populations. COVID-19 vaccines are therefore highly immunogenic in high-risk adolescents and dual vaccination might be able to provide relative protection against the Omicron variant that is currently globally dominant.

Summary: Background: The role of educational settings in SARS-CoV-2 infection and transmission remains controversial. We investigated SARS-CoV-2 infection, seroprevalence, and seroconversion rates in secondary schools during the 2020/21 academic year, which included the emergence of the more transmissible alpha and delta variants, in England. Methods: The UK Health Security Agency (UKHSA) initiated prospective surveillance in 18 urban English secondary schools. Participants had nasal swabs for SARS-CoV-2 RT-PCR and blood sampling for SARS-CoV-2 nucleoprotein and spike protein antibodies at the start (Round 1: September-October 2020) and end (Round 2: December 2020) of the autumn term, when schools reopened after national lockdown was imposed in January 2021 (Round 3: March-April 2021), and end of the academic year (Round 4: May-July 2021). Findings: We enrolled 2314 participants (1277 students, 1037 staff; one participant had missing data for PCR testing). In-school testing identified 31 PCR-positive participants (20 students, 11 staff). Another 247 confirmed cases (112 students, 135 staff) were identified after linkage with national surveillance data, giving an overall positivity rate of 12.0% (278/2313; staff: 14.1%, 146/1037 vs students: 10.3%, 132/1276; p = 0.006). Trends were similar to national infection data. Nucleoprotein-antibody seroprevalence increased for students and staff between Rounds 1 and 3 but were similar between Rounds 3 and 4, when the delta variant was the dominant circulating strain. Overall, Nucleoprotein-antibody seroconversion was 18.4% (137/744) in staff and 18.8% (146/778) in students, while Spike-antibody seroconversion was higher in staff (72.8%, 525/721) than students (21.3%, 163/764) because of vaccination. Interpretation: SARS-CoV-2 infection rates in secondary schools remained low when community infection rates were low, even as the delta variant was emerging in England. Funding: This study was funded by the UK Department of Health and Social Care.

ABSTRACT Seroepidemiological studies to monitor antibody kinetics are important for assessing the extent and spread of SARS-CoV-2 in a population. Noninvasive sampling methods are advantageous for reducing the need for venipuncture, which may be a barrier to investigations, particularly in pediatric populations. Oral fluids are obtained by gingiva-crevicular sampling from children and adults and are very well accepted. Enzyme immunoassays (EIAs) based on these samples have acceptable sensitivity and specificity compared to conventional serum-based antibody EIAs and are suitable for population-based surveillance. We describe the development and evaluation of SARS-CoV-2 IgG EIAs using SARS-CoV-2 viral nucleoprotein (NP) and spike (S) proteins in IgG isotype capture format and an indirect receptor-binding-domain (RBD) IgG EIA, intended for use in children as a primary endpoint. All three assays were assessed using a panel of 1,999 paired serum and oral fluids from children and adults participating in school SARS-CoV-2 surveillance studies during and after the first and second pandemic wave in the United Kingdom. The anti-NP IgG capture assay was the best candidate, with an overall sensitivity of 75% (95% confidence interval [CI]: 71 to 79%) and specificity of 99% (95% CI: 78 to 99%) compared with paired serum antibodies. Sensitivity observed in children (80%, 95% CI: 71 to 88%) was higher than that in adults (67%, CI: 60% to 74%). Oral fluid assays (OF) using spike protein and RBD antigens were also 99% specific and achieved reasonable but lower sensitivity in the target population (78%, 95% CI [68% to 86%] and 53%, 95% CI [43% to 64%], respectively). IMPORTANCE We report on the first large-scale assessment of the suitability of oral fluids for detection of SARS-CoV-2 antibody obtained from healthy children attending school. The sample type (gingiva-crevicular fluid, which is a transudate of blood but is not saliva) can be self collected. Although detection of antibodies in oral fluids is less sensitive than that in blood, our study suggests an optimal format for operational use. The laboratory methods we have developed can reliably measure antibodies in children, who are able to take their own samples. Our findings are of immediate practical relevance for use in large-scale seroprevalence studies designed to measure exposure to infection, as they typically require venipuncture. Overall, our data indicate that OF assays based on the detection of SARS-CoV-2 antibodies are a tool suitable for population-based seroepidemiology studies in children and highly acceptable in children and adults, as venipuncture is no longer necessary.

Background: Older children have higher SARS-CoV-2 infection rates than younger children. We investigated SARS-CoV-2 infection, seroprevalence and seroconversion rates in staff and students following the full reopening of all secondary schools in England. Methods: Public Health England (PHE) invited secondary schools in six regions (East and West London, Hertfordshire, Derbyshire, Manchester and Birmingham) to participate in SARS-CoV-2 surveillance during the 2020/21 academic year. Participants had nasal swabs for RT-PCR and blood samples for SARS-CoV-2 antibodies at the beginning (September 2020) and end (December 2020) of the autumn term. Multivariable logistic regression was used to assess independent risk factors for seropositivity and seroconversion. Findings: Eighteen schools in six regions enrolled 2,209 participants, including 1,189 (53.8%) students and 1,020 (46.2%) staff. SARS-CoV-2 infection rates were not significantly different between students and staff in round one (5/948; [0.53%] vs. 2/876 [0.23%]; p = 0.46) or round two (10/948 [1.05%] vs. 7/886 [0.79%]; p = 0.63), and similar to national prevalence. None of four and 7/15 (47%) sequenced strains in rounds 1 and 2 were the highly transmissible SARS-CoV-2 B.1.1.7 variant. In round 1, antibody seropositivity was higher in students than staff (114/893 [12.8%] vs. 79/861 [9.2%]; p = 0.016), but similar in round 2 (117/893 [13.1%] vs.117/872 [13.3%]; p = 0.85), comparable to local community seroprevalence. Between the two rounds, 8.7% (57/652) staff and 6.6% (36/549) students seroconverted (p = 0.16). Interpretation: In secondary schools, SARS-CoV-2 infection, seropositivity and seroconversion rates were similar in staff and students, and comparable to local community rates. Ongoing surveillance will be important for monitoring the impact of new variants in educational settings.

Recently, host whole blood gene expression signatures have been identified for diagnosis of tuberculosis (TB). Absolute quantification of the concentrations of signature transcripts in blood have not been reported, but would facilitate diagnostic test development. To identify minimal transcript signatures, we applied a transcript selection procedure to microarray data from African adults comprising 536 patients with TB, other diseases (OD) and latent TB (LTBI), divided into training and test sets. Signatures were further investigated using reverse transcriptase (RT)—digital PCR (dPCR). A four-transcript signature (GBP6, TMCC1, PRDM1, and ARG1) measured using RT-dPCR distinguished TB patients from those with OD (area under the curve (AUC) 93.8% (CI95% 82.2–100%). A three-transcript signature (FCGR1A, ZNF296, and C1QB) differentiated TB from LTBI (AUC 97.3%, CI95%: 93.3–100%), regardless of HIV. These signatures have been validated across platforms and across samples offering strong, quantitative support for their use as diagnostic biomarkers for TB.

A correction to this article has been published and is linked from the HTML and PDF versions of this paper. The error has been fixed in the paper.

Tuberculosis (TB) is a leading cause of death globally. Natural history studies show that young children are at particularly high risk of progression to active TB and severe, disseminated disease following infection. Despite this, high-quality regional and global surveillance data on the burden of childhood TB are lacking. We discuss the unique aspects of TB in children that make diagnosis and therefore surveillance challenging; the limitations of available surveillance data; other data which provide insights into the true burden of childhood TB. Improved surveillance is among the key research priorities identified for childhood TB, but progress to date has been slow. Recent advances in TB diagnostics, and standardized clinical diagnostic guidelines and case definitions, all provide opportunities for new strategies to improve surveillance. Better-quality data on the burden and trends of childhood TB will inform and improve both public health policy and clinical practice.

BackgroundSepsis is a common cause of serious illness and death. Sepsis management remains challenging and suboptimal. To support rapid sepsis diagnosis and treatment, screening tools have been embedded into hospital digital systems to appear as digital alerts. The implementation of digital alerts to improve the management of sepsis and deterioration is a complex intervention that has to fit with team workflow and the views and practices of hospital staff. Despite the importance of human decision-making and behavior in optimal implementation, there are limited qualitative studies that explore the views and experiences of health care professionals regarding digital alerts as sepsis or deterioration computerized clinician decision support systems (CCDSSs). ObjectiveThis study aims to explore the views and experiences of health care professionals on the use of sepsis or deterioration CCDSSs and to identify barriers and facilitators to their implementation and use in National Health Service (NHS) hospitals. MethodsWe conducted a qualitative, multisite study with unstructured observations and semistructured interviews with health care professionals from emergency departments, outreach teams, and intensive or acute units in 3 NHS hospital trusts in England. Data from both interviews and observations were analyzed together inductively using thematic analysis. ResultsA total of 22 health care professionals were interviewed, and 12 observation sessions were undertaken. A total of four themes regarding digital alerts were identified: (1) support decision-making as nested in electronic health records, but never substitute professionals’ knowledge and experience; (2) remind to take action according to the context, such as the hospital unit and the job role; (3) improve the alerts and their introduction, by making them more accessible, easy to use, not intrusive, more accurate, as well as integrated across the whole health care system; and (4) contextual factors affecting views and use of alerts in the NHS trusts. Digital alerts are more optimally used in general hospital units with a lower senior decision maker:patient ratio and by health care professionals with experience of a similar technology. Better use of the alerts was associated with quality improvement initiatives and continuous sepsis training. The trusts’ features, such as the presence of a 24/7 emergency outreach team, good technological resources, and staffing and teamwork, favored a more optimal use. ConclusionsTrust implementation of sepsis or deterioration CCDSSs requires support on multiple levels and at all phases of the intervention, starting from a prego-live analysis addressing organizational needs and readiness. Advancements toward minimally disruptive and smart digital alerts as sepsis or deterioration CCDSSs, which are more accurate and specific but at the same time scalable and accessible, require policy changes and investments in multidisciplinary research.

Objectives Describe digital sepsis alerts (DSAs) in use in English National Health Service (NHS) acute hospitals.Methods A Freedom of Information request surveyed acute NHS Trusts on their adoption of electronic patient records (EPRs) and DSAs.Results Of the 99 Trusts that responded, 84 had an EPR. Over 20 different EPR system providers were identified as operational in England. The most common providers were Cerner (21%). System C, Dedalus and Allscripts Sunrise were also relatively common (13%, 10% and 7%, respectively). 70% of NHS Trusts with an EPR responded that they had a DSA; most of these use the National Early Warning Score (NEWS2). There was evidence that the EPR provider was related to the DSA algorithm. We found no evidence that Trusts were using EPRs to introduce data driven algorithms or DSAs able to include, for example, pre-existing conditions that may be known to increase risk.Not all Trusts were willing or able to provide details of their EPR or the underlying algorithm.Discussion The majority of NHS Trusts use an EPR of some kind; many use a NEWS2-based DSA in keeping with national guidelines.Conclusion Many English NHS Trusts use DSAs; even those using similar triggers vary and many recreate paper systems. Despite the proliferation of machine learning algorithms being developed to support early detection of sepsis, there is little evidence that these are being used to improve personalised sepsis detection.

Objective To assess implementation and ease of implementation of control measures in schools as reported by staff and parents.Design A descriptive cross-sectional survey.Setting Staff and parents/guardians of the 132 primary schools and 19 secondary schools participating in COVID-19 surveillance in school kids (sKIDs and sKIDsPLUS Studies).Main outcome measure Prevalence of control measures implemented in schools in autumn 2020, parental and staff perception of ease of implementation.Results In total, 56 of 151 (37%) schools participated in this study, with 1953 parents and 986 staff members completing the questionnaire. Most common measures implemented by schools included regular hand cleaning for students (52 of 56, 93%) and staff (70 of 73, 96%), as reported by parents and staff, respectively, and was among the easiest to implement at all times for students (57%) and even more so, for staff (78%). Maintaining 2-metre distancing was less commonly reported for students (24%–51%) as it was for staff (81%–84%), but was one of the most difficult to follow at all times for students (25%) and staff (16%) alike. Some measures were more commonly reported by primary school compared to secondary school parents, including keeping students within the same small groups (28 of 41, 68% vs 8 of 15, 53%), ensuring the same teacher for classes (29 of 41, 71% vs 6 of 15, 40%). On the other hand, wearing a face covering while at school was reported by three-quarters of secondary school parents compared with only parents of 4 of 41 (10%) primary schools. Other measures such as student temperature checks (5%–13%) and advising staff work from home if otherwise healthy (7%–15%) were rarely reported.Conclusions Variable implementation of infection control measures was reported, with some easier to implement (hand hygiene) than others (physical distancing).

Background: Prospective, longitudinal SARS-CoV-2 sero-surveillance in schools across England was initiated after the first national lockdown, allowing comparison of child and adult antibody responses over time. Methods: Prospective active serological surveillance in 46 primary schools in England tested for SARS-CoV-2 antibodies during June, July and December 2020. Samples were tested for nucleocapsid (N) and receptor binding domain (RBD) antibodies, to estimate antibody persistence at least 6 months after infection, and for the correlation of N, RBD and live virus neutralising activity. Findings: In June 2020, 1,344 staff and 835 students were tested. Overall, 11.5% (95%CI: 9.4–13.9) and 11.3% (95%CI: 9.2–13.6; p = 0.88) of students had nucleoprotein and RBD antibodies, compared to 15.6% (95%CI: 13.7–17.6) and 15.3% (95%CI: 13.4–17.3; p = 0.83) of staff. Live virus neutralising activity was detected in 79.8% (n = 71/89) of nucleocapsid and 85.5% (71/83) of RBD antibody positive children. RBD antibodies correlated more strongly with neutralising antibodies (rs=0.7527; p6 months) but further work is required to understand the extent to which this protects against reinfection.

We conducted voluntary Covid-19 testing programmes for symptomatic and asymptomatic staff at a UK teaching hospital using naso-/oro-pharyngeal PCR testing and immunoassays for IgG antibodies. 1128/10,034 (11.2%) staff had evidence of Covid-19 at some time. Using questionnaire data provided on potential risk-factors, staff with a confirmed household contact were at greatest risk (adjusted odds ratio [aOR] 4.82 [95%CI 3.45–6.72]). Higher rates of Covid-19 were seen in staff working in Covid-19-facing areas (22.6% vs. 8.6% elsewhere) (aOR 2.47 [1.99–3.08]). Controlling for Covid-19-facing status, risks were heterogenous across the hospital, with higher rates in acute medicine (1.52 [1.07–2.16]) and sporadic outbreaks in areas with few or no Covid-19 patients. Covid-19 intensive care unit staff were relatively protected (0.44 [0.28–0.69]), likely by a bundle of PPE-related measures. Positive results were more likely in Black (1.66 [1.25–2.21]) and Asian (1.51 [1.28–1.77]) staff, independent of role or working location, and in porters and cleaners (2.06 [1.34–3.15]).

Abstract Background Prosthetic joint infections are clinically difficult to diagnose and treat. Previously, we demonstrated metagenomic sequencing on an Illumina MiSeq replicates the findings of current gold standard microbiological diagnostic techniques. Nanopore sequencing offers advantages in speed of detection over MiSeq. Here, we report a real-time analytical pathway for Nanopore sequence data, designed for detecting bacterial composition of prosthetic joint infections but potentially useful for any microbial sequencing, and compare detection by direct-from-clinical-sample metagenomic nanopore sequencing with Illumina sequencing and standard microbiological diagnostic techniques. Results DNA was extracted from the sonication fluids of seven explanted orthopaedic devices, and additionally from two culture negative controls, and was sequenced on the Oxford Nanopore Technologies MinION platform. A specific analysis pipeline was assembled to overcome the challenges of identifying the true infecting pathogen, given high levels of host contamination and unavoidable background lab and kit contamination. The majority of DNA classified (> 90%) was host contamination and discarded. Using negative control filtering thresholds, the species identified corresponded with both routine microbiological diagnosis and MiSeq results. By analysing sequences in real time, causes of infection were robustly detected within minutes from initiation of sequencing. Conclusions We demonstrate a novel, scalable pipeline for real-time analysis of MinION sequence data and use of this pipeline to show initial proof of concept that metagenomic MinION sequencing can provide rapid, accurate diagnosis for prosthetic joint infections. The high proportion of human DNA in prosthetic joint infection extracts prevents full genome analysis from complete coverage, and methods to reduce this could increase genome depth and allow antimicrobial resistance profiling. The nine samples sequenced in this pilot study have shown a proof of concept for sequencing and analysis that will enable us to investigate further sequencing to improve specificity and sensitivity.

BackgroundA major impediment to tuberculosis control in Africa is the difficulty in diagnosing active tuberculosis (TB), particularly in the context of HIV infection. We hypothesized that a unique host blood RNA transcriptional signature would distinguish TB from other diseases (OD) in HIV-infected and -uninfected patients, and that this could be the basis of a simple diagnostic test.Methods and findingsAdult case-control cohorts were established in South Africa and Malawi of HIV-infected or -uninfected individuals consisting of 584 patients with either TB (confirmed by culture of Mycobacterium tuberculosis [M.TB] from sputum or tissue sample in a patient under investigation for TB), OD (i.e., TB was considered in the differential diagnosis but then excluded), or healthy individuals with latent TB infection (LTBI). Individuals were randomized into training (80%) and test (20%) cohorts. Blood transcriptional profiles were assessed and minimal sets of significantly differentially expressed transcripts distinguishing TB from LTBI and OD were identified in the training cohort. A 27 transcript signature distinguished TB from LTBI and a 44 transcript signature distinguished TB from OD. To evaluate our signatures, we used a novel computational method to calculate a disease risk score (DRS) for each patient. The classification based on this score was first evaluated in the test cohort, and then validated in an independent publically available dataset (GSE19491). In our test cohort, the DRS classified TB from LTBI (sensitivity 95%, 95% CI [87-100]; specificity 90%, 95% CI [80-97]) and TB from OD (sensitivity 93%, 95% CI [83-100]; specificity 88%, 95% CI [74-97]). In the independent validation cohort, TB patients were distinguished both from LTBI individuals (sensitivity 95%, 95% CI [85-100]; specificity 94%, 95% CI [84-100]) and OD patients (sensitivity 100%, 95% CI [100-100]; specificity 96%, 95% CI [93-100]). Limitations of our study include the use of only culture confirmed TB patients, and the potential that TB may have been misdiagnosed in a small proportion of OD patients despite the extensive clinical investigation used to assign each patient to their diagnostic group.ConclusionsIn our study, blood transcriptional signatures distinguished TB from other conditions prevalent in HIV-infected and -uninfected African adults. Our DRS, based on these signatures, could be developed as a test for TB suitable for use in HIV endemic countries. Further evaluation of the performance of the signatures and DRS in prospective populations of patients with symptoms consistent with TB will be needed to define their clinical value under operational conditions. Please see later in the article for the Editors' Summary.