Inference in High-Dimensional Online Changepoint Detection.

We introduce and study two new inferential challenges associated with the sequential detection of change in a high-dimensional mean vector. First, we seek a confidence interval for the changepoint, and second, we estimate the set of indices of coordinates in which the mean changes. We propose an online algorithm that produces an interval with guaranteed nominal coverage, and whose length is, with high probability, of the same order as the average detection delay, up to a logarithmic factor. The corresponding support estimate enjoys control of both false negatives and false positives. Simulations confirm the effectiveness of our methodology, and we also illustrate its applicability on the U.S. excess deaths data from 2017 to 2020. The supplementary material, which contains the proofs of our theoretical results, is available online.

Career intentions of medical students in the UK: a national, cross-sectional study (AIMS study).

OBJECTIVE: To determine current UK medical students' career intentions after graduation and on completing the Foundation Programme (FP), and to ascertain the motivations behind these intentions. DESIGN: Cross-sectional, mixed-methods survey of UK medical students, using a non-random sampling method. SETTING: All 44 UK medical schools recognised by the General Medical Council. PARTICIPANTS: All UK medical students were eligible to participate. The study sample consisted of 10 486 participants, approximately 25.50% of the medical student population. OUTCOME MEASURES: Career intentions of medical students postgraduation and post-FP, motivations behind these career intentions, characterising the medical student population and correlating demographic factors and propensity to leave the National Health Service (NHS). RESULTS: The majority of participating students (8806/10 486, 83.98%) planned to complete both years of the FP after graduation, with under half of these students (4294/8806, 48.76%) intending to pursue specialty training thereafter. A subanalysis of career intentions after the FP by year of study revealed a significant decrease in students' intentions to enter specialty training as they advanced through medical school. Approximately a third of surveyed students (3392/10 486, 32.35%) intended to emigrate to practise medicine, with 42.57% (n=1444) of those students not planning to return. In total, 2.89% of students intended to leave medicine altogether (n=303). Remuneration, work-life balance and working conditions were identified as important factors in decision-making regarding emigration and leaving the profession. Subgroup analyses based on gender, type of schooling, fee type and educational background were performed. Only 17.26% of surveyed students were satisfied or very satisfied with the overall prospect of working in the NHS. CONCLUSIONS: The Ascertaining the career Intentions of UK Medical Students study highlights UK students' views and career intentions, revealing a concerning proportion of those surveyed considering alternative careers or emigration. Addressing factors such as remuneration, work-life balance and working conditions may increase retention of doctors and improve workforce planning efforts.

High-dimensional principal component analysis with heterogeneous missingness.

We study the problem of high-dimensional Principal Component Analysis (PCA) with missing observations. In a simple, homogeneous observation model, we show that an existing observed-proportion weighted (OPW) estimator of the leading principal components can (nearly) attain the minimax optimal rate of convergence, which exhibits an interesting phase transition. However, deeper investigation reveals that, particularly in more realistic settings where the observation probabilities are heterogeneous, the empirical performance of the OPW estimator can be unsatisfactory; moreover, in the noiseless case, it fails to provide exact recovery of the principal components. Our main contribution, then, is to introduce a new method, which we call primePCA, that is designed to cope with situations where observations may be missing in a heterogeneous manner. Starting from the OPW estimator, primePCA iteratively projects the observed entries of the data matrix onto the column space of our current estimate to impute the missing entries, and then updates our estimate by computing the leading right singular space of the imputed data matrix. We prove that the error of primePCA converges to zero at a geometric rate in the noiseless case, and when the signal strength is not too small. An important feature of our theoretical guarantees is that they depend on average, as opposed to worst-case, properties of the missingness mechanism. Our numerical studies on both simulated and real data reveal that primePCA exhibits very encouraging performance across a wide range of scenarios, including settings where the data are not Missing Completely At Random.

Single-dose BNT162b2 vaccine protects against asymptomatic SARS-CoV-2 infection.

The BNT162b2 mRNA COVID-19 vaccine (Pfizer-BioNTech) is being utilised internationally for mass COVID-19 vaccination. Evidence of single-dose protection against symptomatic disease has encouraged some countries to opt for delayed booster doses of BNT162b2, but the effect of this strategy on rates of asymptomatic SARS-CoV-2 infection remains unknown. We previously demonstrated frequent pauci- and asymptomatic SARS-CoV-2 infection amongst healthcare workers (HCWs) during the UK's first wave of the COVID-19 pandemic, using a comprehensive PCR-based HCW screening programme (Rivett et al., 2020; Jones et al., 2020). Here, we evaluate the effect of first-dose BNT162b2 vaccination on test positivity rates and find a fourfold reduction in asymptomatic infection amongst HCWs ≥12 days post-vaccination. These data provide real-world evidence of short-term protection against asymptomatic SARS-CoV-2 infection following a single dose of BNT162b2 vaccine, suggesting that mass first-dose vaccination will reduce SARS-CoV-2 transmission, as well as the burden of COVID-19 disease.

USP: an independence test that improves on Pearson's chi-squared and the G-test.

We present the U -statistic permutation (USP) test of independence in the context of discrete data displayed in a contingency table. Either Pearson's χ 2 -test of independence, or the G -test, are typically used for this task, but we argue that these tests have serious deficiencies, both in terms of their inability to control the size of the test, and their power properties. By contrast, the USP test is guaranteed to control the size of the test at the nominal level for all sample sizes, has no issues with small (or zero) cell counts, and is able to detect distributions that violate independence in only a minimal way. The test statistic is derived from a U -statistic estimator of a natural population measure of dependence, and we prove that this is the unique minimum variance unbiased estimator of this population quantity. The practical utility of the USP test is demonstrated on both simulated data, where its power can be dramatically greater than those of Pearson's test, the G -test and Fisher's exact test, and on real data. The USP test is implemented in the R package USP.

Screening of healthcare workers for SARS-CoV-2 highlights the role of asymptomatic carriage in COVID-19 transmission.

Significant differences exist in the availability of healthcare worker (HCW) SARS-CoV-2 testing between countries, and existing programmes focus on screening symptomatic rather than asymptomatic staff. Over a 3 week period (April 2020), 1032 asymptomatic HCWs were screened for SARS-CoV-2 in a large UK teaching hospital. Symptomatic staff and symptomatic household contacts were additionally tested. Real-time RT-PCR was used to detect viral RNA from a throat+nose self-swab. 3% of HCWs in the asymptomatic screening group tested positive for SARS-CoV-2. 17/30 (57%) were truly asymptomatic/pauci-symptomatic. 12/30 (40%) had experienced symptoms compatible with coronavirus disease 2019 (COVID-19)>7 days prior to testing, most self-isolating, returning well. Clusters of HCW infection were discovered on two independent wards. Viral genome sequencing showed that the majority of HCWs had the dominant lineage B∙1. Our data demonstrates the utility of comprehensive screening of HCWs with minimal or no symptoms. This approach will be critical for protecting patients and hospital staff.

Patients admitted to NHS hospitals are now routinely screened for SARS-CoV-2 (the virus that causes COVID-19), and isolated from other patients if necessary. Yet healthcare workers, including frontline patient-facing staff such as doctors, nurses and physiotherapists, are only tested and excluded from work if they develop symptoms of the illness. However, there is emerging evidence that many people infected with SARS-CoV-2 never develop significant symptoms: these people will therefore be missed by 'symptomatic-only' testing. There is also important data showing that around half of all transmissions of SARS-CoV-2 happen before the infected individual even develops symptoms. This means that much broader testing programs are required to spot people when they are most infectious. Rivett, Sridhar, Sparkes, Routledge et al. set out to determine what proportion of healthcare workers was infected with SARS-CoV-2 while also feeling generally healthy at the time of testing. Over 1,000 staff members at a large UK hospital who felt they were well enough to work, and did not fit the government criteria for COVID-19 infection, were tested. Amongst these, 3% were positive for SARS-CoV-2. On closer questioning, around one in five reported no symptoms, two in five very mild symptoms that they had dismissed as inconsequential, and a further two in five reported COVID-19 symptoms that had stopped more than a week previously. In parallel, healthcare workers with symptoms of COVID-19 (and their household contacts) who were self-isolating were also tested, in order to allow those without the virus to quickly return to work and bolster a stretched workforce. Finally, the rates of infection were examined to probe how the virus could have spread through the hospital and among staff ­ and in particular, to understand whether rates of infection were greater among staff working in areas devoted to COVID-19 patients. Despite wearing appropriate personal protective equipment, healthcare workers in these areas were almost three times more likely to test positive than those working in areas without COVID-19 patients. However, it is not clear whether this genuinely reflects greater rates of patients passing the infection to staff. Staff may give the virus to each other, or even acquire it at home. Overall, this work implies that hospitals need to be vigilant and introduce broad screening programmes across their workforces. It will be vital to establish such approaches before 'lockdown' is fully lifted, so healthcare institutions are prepared for any second peak of infections.

Multicentre study of physical abuse and limb fractures in young children in the East Anglia Region, UK.

OBJECTIVE: To determine if the detection of physical abuse in young children with fractures is of uniform high standard in the East Anglia Region of the UK, and whether we can identify areas for improvement in our detection of high-risk groups. DESIGN: Multicentre retrospective 4-year study. SETTING: 7 hospitals across the East Anglia Region of Britain (East Anglia Paediatric Physical Abuse and Fractures study). PARTICIPANTS: Age groups and fractures indicated as being at higher risk for physical abuse (all children under 12 months of age, and fractures of humerus and femur in children under 36 months of age). OUTCOME MEASURES: Our criterion for physical abuse was the decision of a multiagency child protection case conference (CPCC). RESULTS: Probability of CPCC decision of physical abuse was highest in infants, ranging from 50% of fractures sustained in the first month of life (excluding obstetric injuries) to 10% at 12 months of age. Only 46%-86% of infants (under 12 months) with a fracture were assessed by a paediatrician for physical abuse after their fracture. Significant variation in the use of skeletal surveys and in CPCC decision of physical abuse was noted in children attending different hospitals. CONCLUSIONS: It is a concern that significant variation between hospitals was found in the investigation and detection of physical abuse as confirmed by CPCC decisions. To minimise failure to detect true cases of physical abuse, we recommend that all high-risk children should be assessed by a paediatrician prior to discharge from the emergency department. Our proposed criteria for assessment (where we found probability of CPCC decision of physical abuse was at least 10%) are any child under the age of 12 months with any fracture, under 18 months of age with femur fracture and under 24 months with humeral shaft fracture (not supracondylar).

A multiple myeloma classification system that associates normal B-cell subset phenotypes with prognosis.

Despite the recent progress in treatment of multiple myeloma (MM), it is still an incurable malignant disease, and we are therefore in need of new risk stratification tools that can help us to understand the disease and optimize therapy. Here we propose a new subtyping of myeloma plasma cells (PCs) from diagnostic samples, assigned by normal B-cell subset associated gene signatures (BAGS). For this purpose, we combined fluorescence-activated cell sorting and gene expression profiles from normal bone marrow (BM) Pre-BI, Pre-BII, immature, naïve, memory, and PC subsets to generate BAGS for assignment of normal BM subtypes in diagnostic samples. The impact of the subtypes was analyzed in 8 available data sets from 1772 patients' myeloma PC samples. The resulting tumor assignments in available clinical data sets exhibited similar BAGS subtype frequencies in 4 cohorts from de novo MM patients across 1296 individual cases. The BAGS subtypes were significantly associated with progression-free and overall survival in a meta-analysis of 916 patients from 3 prospective clinical trials. The major impact was observed within the Pre-BII and memory subtypes, which had a significantly inferior prognosis compared with other subtypes. A multiple Cox proportional hazard analysis documented that BAGS subtypes added significant, independent prognostic information to the translocations and cyclin D classification. BAGS subtype analysis of patient cases identified transcriptional differences, including a number of differentially spliced genes. We identified subtype differences in myeloma at diagnosis, with prognostic impact and predictive potential, supporting an acquired B-cell trait and phenotypic plasticity as a pathogenetic hallmark of MM.

Diffuse large B-cell lymphoma classification system that associates normal B-cell subset phenotypes with prognosis.

PURPOSE: Current diagnostic tests for diffuse large B-cell lymphoma use the updated WHO criteria based on biologic, morphologic, and clinical heterogeneity. We propose a refined classification system based on subset-specific B-cell-associated gene signatures (BAGS) in the normal B-cell hierarchy, hypothesizing that it can provide new biologic insight and diagnostic and prognostic value. PATIENTS AND METHODS: We combined fluorescence-activated cell sorting, gene expression profiling, and statistical modeling to generate BAGS for naive, centrocyte, centroblast, memory, and plasmablast B cells from normal human tonsils. The impact of BAGS-assigned subtyping was analyzed using five clinical cohorts (treated with cyclophosphamide, doxorubicin, vincristine, and prednisone [CHOP], n = 270; treated with rituximab plus CHOP [R-CHOP], n = 869) gathered across geographic regions, time eras, and sampling methods. The analysis estimated subtype frequencies and drug-specific resistance and included a prognostic meta-analysis of patients treated with first-line R-CHOP therapy. RESULTS: Similar BAGS subtype frequencies were assigned across 1,139 samples from five different cohorts. Among R-CHOP-treated patients, BAGS assignment was significantly associated with overall survival and progression-free survival within the germinal center B-cell-like subclass; the centrocyte subtype had a superior prognosis compared with the centroblast subtype. In agreement with the observed therapeutic outcome, centrocyte subtypes were estimated as being less resistant than the centroblast subtype to doxorubicin and vincristine. The centroblast subtype had a complex genotype, whereas the centrocyte subtype had high TP53 mutation and insertion/deletion frequencies and expressed LMO2, CD58, and stromal-1-signature and major histocompatibility complex class II-signature genes, which are known to have a positive impact on prognosis. CONCLUSION: Further development of a diagnostic platform using BAGS-assigned subtypes may allow pathogenetic studies to improve disease management.