AbstractBackgroundNon-pharmacological interventions were introduced based on modelling studies which suggested that the English National Health Service (NHS) would be overwhelmed by the COVID-19 pandemic. In this study, we describe the pattern of bed occupancy across England during the first wave of the pandemic, January 31st to June 5th 2020.MethodsBed availability and occupancy data was extracted from daily reports submitted by all English secondary care providers, between 27-Mar and 5-June. Two thresholds for ‘safe occupancy’ were utilized (85% as per Royal College of Emergency Medicine and 92% as per NHS Improvement).FindingsAt peak availability, there were 2711 additional beds compatible with mechanical ventilation across England, reflecting a 53% increase in capacity, and occupancy never exceeded 62%. A consequence of the repurposing of beds meant that at the trough, there were 8·7% (8,508) fewer general and acute (G&A) beds across England, but occupancy never exceeded 72%. The closest to (surge) capacity that any trust in England reached was 99·8% for general and acute beds. For beds compatible with mechanical ventilation there were 326 trust-days (3·7%) spent above 85% of surge capacity, and 154 trust-days (1·8%) spent above 92%. 23 trusts spent a cumulative 81 days at 100% saturation of their surge ventilator bed capacity (median number of days per trust = 1 [range: 1 to 17]). However, only 3 STPs (aggregates of geographically co-located trusts) reached 100% saturation of their mechanical ventilation beds.InterpretationThroughout the first wave of the pandemic, an adequate supply of all bed-types existed at a national level. Due to an unequal distribution of bed utilization, many trusts spent a significant period operating above ‘safe-occupancy’ thresholds, despite substantial capacity in geographically co-located trusts; a key operational issue to address in preparing for a potential second wave.FundingThis study received no funding.Research in ContextEvidence Before This StudyWe identified information sources describing COVID-19 related bed and mechanical ventilator demand modelling, as well as bed occupancy during the first wave of the pandemic by performing regular searches of MedRxiv, PubMed and Google, using the terms ‘COVID-19’, ‘mechanical ventilators’, ‘bed occupancy’, ‘England’, ‘UK’, ‘demand’, and ‘non-pharmacological interventions (NPIs)’, until June 20th, 2020. Two UK-specific studies were found that modelled the demand for mechanical ventilators, one of which incorporated sensitivity analysis based on the introduction of NPIs and found that their effects might prevent the healthcare system being overwhelmed. Separately, several news reports were found pertaining to a single hospital that reached ventilator capacity in England during the first wave of the pandemic, however, no single authoritative source was identified detailing impact across all hospital sites in England.Added Value of This StudyThis national study of hospital-level bed occupancy in England provides unique and timely insight into bed-specific resource utilization during the first wave of the COVID-19 pandemic, nationally, and by specific (geographically defined) health footprints. We found evidence of an unequal distribution of resource utilization across England. Although occupancy of beds compatible with mechanical ventilation never exceeded 62% at the national level, 52 (30%) hospitals across England reached 100% saturation at some point during the first wave of the pandemic. Close examination of the geospatial data revealed that in the vast majority of circumstances there was relief capacity in geographically co-located hospitals. Over the first wave it was theoretically possible to markedly reduce (by 95.1%) the number of hospitals at 100% saturation of their mechanical ventilator bed capacity by redistributing patients to nearby hospitals.Implications of all the Available EvidenceNow-casting using routinely collected administrative data presents a robust approach to rapidly evaluate the effectiveness of national policies introduced to prevent a healthcare system being overwhelmed in the context of a pandemic illness. Early investment in operational field hospital and an independent sector network may yield more overtly positive results in the winter, when G&A occupancy-levels regularly exceed 92% in England, however, during the first wave of the pandemic they were under-utilized. Moreover, in the context of the non-pharmacological interventions utilized during the first wave of COVID-19, demand for beds and mechanical ventilators was much lower than initially predicted, but despite this many trust spent a significant period of time operating above ‘safe-occupancy’ thresholds. This finding demonstrates that it is vital that future demand (prediction) models reflect the nuances of local variation within a healthcare system. Failure to incorporate such geographical variation can misrepresent the likelihood of surpassing availability thresholds by averaging out over regions with relatively lower demand, and presents a key operational issue for policymakers to address in preparing for a potential second wave.

Objective: Precision medicine requires reliable identification of variation in patient-level outcomes with different available treatments, often termed treatment effect heterogeneity. We aimed to evaluate the comparative utility of individualized treatment selection strategies based on predicted individual-level treatment effects from a causal forest machine learning algorithm and a penalized regression model.

Methods: Cohort study characterizing individual-level glucose-lowering response (6 month reduction in HbA1c) in people with type 2 diabetes initiating SGLT2-inhibitor or DPP4-inhibitor therapy. Model development set comprised 1,428 participants in the CANTATA-D and CANTATA-D2 randomised clinical trials of SGLT2-inhibitors versus DPP4-inhibitors. For external validation, calibration of observed versus predicted differences in HbA1c in patient strata defined by size of predicted HbA1c benefit was evaluated in 18,741 patients in UK primary care (Clinical Practice Research Datalink).

Results: Heterogeneity in treatment effects was detected in clinical trial participants with both approaches (proportion predicted to have a benefit on SGLT2-inhibitor therapy over DPP4-inhibitor therapy: causal forest: 98.6%; penalized regression: 81.7%). In validation, calibration was good with penalized regression but sub-optimal with causal forest. A strata with an HbA1c benefit >10 mmol/mol with SGLT2-inhibitors (3.7% of patients, observed benefit 11.0 mmol/mol [95%CI 8.0-14.0]) was identified using penalized regression but not causal forest, and a much larger strata with an HbA1c benefit 5-10 mmol with SGLT2-inhibitors was identified with penalized regression (regression: 20.9% of patients, observed benefit 7.8 mmol/mol (95%CI 6.7-8.9); causal forest 11.6%, observed benefit 8.7 mmol/mol (95%CI 7.4-10.1).

Conclusions: Consistent with recent results for outcome prediction with clinical data, when evaluating treatment effect heterogeneity researchers should not rely on causal forest or other similar machine learning algorithms alone, and must compare outputs with standard regression, which in this evaluation was superior.

AbstractObjectiveTo establish whether clinical patient characteristics routinely measured in primary care can identify people with differing short-term benefits and risks for SGLT2-inhibitor and DPP4-inhibitor therapies, and to derive and validate a treatment selection algorithm to identify the likely optimal therapy for individual patients.DesignProspective cohort study.SettingRoutine clinical data from United Kingdom general practice (Clinical Practice Research Datalink [CPRD]), and individual-level clinical trial data from 14 multi-country trials of SGLT2-inhibitor and DPP4-inhibitor therapies.Participants26,877 new users of SGLT2-inhibitor and DPP4-inhibitor therapy in CPRD over 2013-2019, and 10,414 participants randomised to SGLT2-inhibitor or DPP4-inhibitor therapy in 14 clinical trials, including 3 head-to-head trials of the two therapies (n=2,499).Main outcome measuresThe primary outcome was achieved HbA1c 6 months after initiating therapy. Clinical features associated with differential HbA1c outcomes with SGLT2-inhibitor and DPP4-inhibitor therapies were identified in routine clinical data, with associations then tested in trial data. A multivariable treatment selection algorithm to predict differential HbA1c outcomes was developed in a CPRD derivation cohort (n=14,069), with validation in a CPRD validation cohort (n=9,376) and the head-to-head trials. In CPRD, we further explored the relationship between model predictions and secondary outcomes of weight loss and treatment discontinuation.ResultsThe final treatment selection algorithm included HbA1c, eGFR, ALT, age, and BMI, which were identified as predictors of differential HbA1c outcomes with SGLT2-inhibitor and DPP4-inhibitor therapies using both routine and trial data. In validation cohorts, patient strata predicted to have a ≥5 mmol/mol HbA1c reduction with SGLT2-inhibitor therapy compared with DPP4-inhibitor therapy (38.8% of CPRD validation sample) had an observed greater reduction of 8.8 mmol/mol [95%CI 7.8-9.8] in the CPRD validation sample, a 5.8 mmol/mol (95%CI 3.9-7.7) greater reduction in the Cantata D/D2 trials, and a 6.6 mmol/mol [95%CI 2.2-11.0]) greater reduction in the BI1245.20 trial. In CPRD, there was a greater weight reduction with SGLT2-inhibitor therapy regardless of predicted glycaemic benefit. Strata predicted to have greater reduction in HbA1c on SGLT2-inhibitor therapy had a similar risk of discontinuation as on DPP4-inhibitor therapy. In contrast, strata predicted to have greater reduction in HbA1c with DPP4-inhibitor therapy were half as likely to discontinue DPP4-inhibitor therapy than SGLT2-inhibitor therapy.ConclusionsRoutinely measured clinical features are robustly associated with differential glycaemic responses to SGLT2-inhibitor and DPP4-inhibitor therapies. Combining features into a treatment selection algorithm can inform clinical decisions concerning optimal type 2 diabetes treatment choices.Key messagesWhat is already known on this subjectDespite there being multiple glucose-lowering treatment options available for people with type 2 diabetes, current guidelines do not provide clear advice on selecting the optimal treatment for most patients.It is unknown whether routinely measured clinical features modify the risks and benefits of two common treatment options, DPP4-inhibitor or SGLT2-inhibitor therapy, and which could be used to target these treatments to those patients most likely to benefit.What this study addsUsing data from 10,414 participants in 14 randomised trials, and 26,877 patients in UK primary care, we show several routinely available clinical features, notably glycated haemoglobin (HbA1c) and kidney function, are robustly associated with differential HbA1c responses to initiating SGLT2-inhibitor and DPP4-inhibitor therapies.Combining clinical features into a multivariable treatment selection model identifies validated patient strata with 1) a >5 mmol/mol HbA1c benefit for SGLT2-i therapy compared with DPP4-inhibitor therapy ; 2) a 50% reduced risk of early treatment discontinuation with DPP4-inhibitor therapy compared with SGLT2-inhibitor therapy.Our findings demonstrate a precision medicine approach based on routine clinical features can inform clinical decisions concerning optimal type 2 diabetes treatment choices.

AbstractAimsPopulation datasets are increasingly used to study type 1 or 2 diabetes, and inform clinical practice. However, correctly classifying diabetes type, when insulin treated, in population datasets is challenging. Many different approaches have been proposed, ranging from simple age or BMI cut offs, to complex algorithms, and the optimal approach is unclear. We aimed to compare the performance of approaches for classifying insulin treated diabetes for research studies, evaluated against two independent biological definitions of diabetes type.MethodWe compared accuracy of thirteen reported approaches for classifying insulin treated diabetes into type 1 and type 2 diabetes in two population cohorts with diabetes: UK Biobank (UKBB) n=26,399 and DARE n=1,296. Overall accuracy and predictive values for classifying type 1 and 2 diabetes were assessed using: 1) a type 1 diabetes genetic risk score and genetic stratification method (UKBB); 2) C-peptide measured at >3 years diabetes duration (DARE).ResultsAccuracy of approaches ranged from 71%-88% in UKBB and 68%-88% in DARE. All approaches were improved by combining with requirement for early insulin treatment (<1 year from diagnosis). When classifying all participants, combining early insulin requirement with a type 1 diabetes probability model incorporating continuous clinical features (diagnosis age and BMI only) consistently achieved high accuracy, (UKBB 87%, DARE 85%). Self-reported diabetes type alone had high accuracy (UKBB 87%, DARE 88%) but was available in just 15% of UKBB participants. For identifying type 1 diabetes with minimal misclassification, using models with high thresholds or young age at diagnosis (<20 years) had the highest performance. An online tool developed from all UKBB findings allows the optimum approach of those tested to be selected based on variable availability and the research aim.ConclusionSelf-reported diagnosis and models combining continuous features with early insulin requirement are the most accurate methods of classifying insulin treated diabetes in research datasets without measured classification biomarkers.

AbstractObjectivesTo determine the trend in mortality risk over time in people with severe COVID-19 requiring critical care (high intensive unit [HDU] or intensive care unit [ICU]) management.MethodsWe accessed national English data on all adult COVID-19 specific critical care admissions from the COVID-19 Hospitalisation in England Surveillance System (CHESS), up to the 29th June 2020 (n=14,958). The study period was 1st March until 30th May, meaning every patient had 30 days of potential follow-up available. The primary outcome was in-hospital 30-day all-cause mortality. Hazard ratios for mortality were estimated for those admitted each week using a Cox proportional hazards models, adjusting for age (non-linear restricted cubic spline), sex, ethnicity, comorbidities, and geographical region.Results30-day mortality peaked for people admitted to critical care in early April (peak 29.1% for HDU, 41.5% for ICU). There was subsequently a sustained decrease in mortality risk until the end of the study period. As a linear trend from the first week of April, adjusted mortality risk decreased by 11.2% (adjusted HR 0.89 [95% CI 0.87 - 0.91]) per week in HDU, and 9.0% (adjusted HR 0.91 [95% CI 0.88 - 0.94]) in ICU.ConclusionsThere has been a substantial mortality improvement in people admitted to critical care with COVID-19 in England, with markedly lower mortality in people admitted in mid-April and May compared to earlier in the pandemic. This trend remains after adjustment for patient demographics and comorbidities suggesting this improvement is not due to changing patient characteristics. Possible causes include the introduction of effective treatments as part of clinical trials and a falling critical care burden.

Background: Updatable estimates of COVID-19 onset, progression, and trajectories underpin pandemic mitigation efforts. To identify and characterise disease trajectories, we aimed to define and validate ten COVID-19 phenotypes from nationwide linked electronic health records (EHR) using an extensible framework. Methods: in this cohort study, we used eight linked National Health Service (NHS) datasets for people in England alive on Jan 23, 2020. Data on COVID-19 testing, vaccination, primary and secondary care records, and death registrations were collected until Nov 30, 2021. We defined ten COVID-19 phenotypes reflecting clinically relevant stages of disease severity and encompassing five categories: positive SARS-CoV-2 test, primary care diagnosis, hospital admission, ventilation modality (four phenotypes), and death (three phenotypes). We constructed patient trajectories illustrating transition frequency and duration between phenotypes. Analyses were stratified by pandemic waves and vaccination status. Findings: Among 57 032 174 individuals included in the cohort, 13 990 423 COVID-19 events were identified in 7 244 925 individuals, equating to an infection rate of 12·7% during the study period. of 7 244 925 individuals, 460 737 (6·4%) were admitted to hospital and 158 020 (2·2%) died. of 460 737 individuals who were admitted to hospital, 48 847 (10·6%) were admitted to the intensive care unit (ICU), 69 090 (15·0%) received non-invasive ventilation, and 25 928 (5·6%) received invasive ventilation. Among 384 135 patients who were admitted to hospital but did not require ventilation, mortality was higher in wave 1 (23 485 [30·4%] of 77 202 patients) than wave 2 (44 220 [23·1%] of 191 528 patients), but remained unchanged for patients admitted to the ICU. Mortality was highest among patients who received ventilatory support outside of the ICU in wave 1 (2569 [50·7%] of 5063 patients). 15 486 (9·8%) of 158 020 COVID-19-related deaths occurred within 28 days of the first COVID-19 event without a COVID-19 diagnoses on the death certificate. 10 884 (6·9%) of 158 020 deaths were identified exclusively from mortality data with no previous COVID-19 phenotype recorded. We observed longer patient trajectories in wave 2 than wave 1. Interpretation: Our analyses illustrate the wide spectrum of disease trajectories as shown by differences in incidence, survival, and clinical pathways. We have provided a modular analytical framework that can be used to monitor the impact of the pandemic and generate evidence of clinical and policy relevance using multiple EHR sources. Funding: British Heart Foundation Data Science Centre, led by Health Data Research UK.

Type 2 diabetes (T2D) is a complex chronic disease characterized by considerable phenotypic heterogeneity. In this study, we applied a reverse graph embedding method to routinely collected data from 23,137 Scottish patients with newly diagnosed diabetes to visualize this heterogeneity and used partitioned diabetes polygenic risk scores to gain insight into the underlying biological processes. Overlaying risk of progression to outcomes of insulin requirement, chronic kidney disease, referable diabetic retinopathy and major adverse cardiovascular events, we show how these risks differ by patient phenotype. For example, patients at risk of retinopathy are phenotypically different from those at risk of cardiovascular events. We replicated our findings in the UK Biobank and the ADOPT clinical trial, also showing that the pattern of diabetes drug monotherapy response differs for different drugs. Overall, our analysis highlights how, in a European population, underlying phenotypic variation drives T2D onset and affects subsequent diabetes outcomes and drug response, demonstrating the need to incorporate these factors into personalized treatment approaches for the management of T2D.

Abstract
. Aims/hypothesis
. Screening programmes can detect cases of undiagnosed diabetes earlier than symptomatic or incidental diagnosis. However, the improvement in time to diagnosis achieved by screening programmes compared with routine clinical care is unclear. We aimed to use the UK Biobank population-based study to provide the first population-based estimate of the reduction in time to diabetes diagnosis that could be achieved by HbA1c-based screening in middle-aged adults.
.
. Methods
. We studied UK Biobank participants aged 40–70 years with HbA1c measured at enrolment (but not fed back to participants/clinicians) and linked primary and secondary healthcare data (n=179,923) and identified those with a pre-existing diabetes diagnosis (n=13,077, 7.3%). Among the remaining participants (n=166,846) without a diabetes diagnosis, we used an elevated enrolment HbA1c level (≥48 mmol/mol [≥6.5%]) to identify those with undiagnosed diabetes. For this group, we used Kaplan–Meier analysis to assess the time between enrolment HbA1c measurement and subsequent clinical diabetes diagnosis up to 10 years, and Cox regression to identify clinical factors associated with delayed diabetes diagnosis.
.
. Results
. In total, 1.0% (1703/166,846) of participants without a diabetes diagnosis had undiagnosed diabetes based on calibrated HbA1c levels at UK Biobank enrolment, with a median HbA1c level of 51.3 mmol/mol (IQR 49.1–57.2) (6.8% [6.6–7.4]). These participants represented an additional 13.0% of diabetes cases in the study population relative to the 13,077 participants with a diabetes diagnosis. The median time to clinical diagnosis for those with undiagnosed diabetes was 2.2 years, with a median HbA1c at clinical diagnosis of 58.2 mmol/mol (IQR 51.0–80.0) (7.5% [6.8–9.5]). Female participants with lower HbA1c and BMI measurements at enrolment experienced the longest delay to clinical diagnosis.
.
. Conclusions/interpretation
. Our population-based study shows that HbA1c screening in adults aged 40–70 years can reduce the time to diabetes diagnosis by a median of 2.2 years compared with routine clinical care. The findings support the use of HbA1c screening to reduce the time for which individuals are living with undiagnosed diabetes.
.
. Graphical abstract
.
.

Objectives
To describe the relationship between reported serious operational problems (SOPs), and mortality for patients with COVID-19 admitted to intensive care units (ICUs).


Design
English national retrospective cohort study.


Setting
89 English hospital trusts (i.e. small groups of hospitals functioning as single operational units).


Patients
All adults with COVID-19 admitted to ICU between 2nd April and 1st December, 2020 (n = 6,737).


Interventions
N/A


Main outcomes and measures
Hospital trusts routinely submit declarations of whether they have experienced ‘serious operational problems’ in the last 24 hours (e.g. due to staffing issues, adverse weather conditions, etc.). Bayesian hierarchical models were used to estimate the association between in-hospital mortality (binary outcome) and: 1) an indicator for whether a SOP occurred on the date of a patient’s admission, and; 2) the proportion of the days in a patient’s stay that had a SOP occur within their trust. These models were adjusted for individual demographic characteristics (age, sex, ethnicity), and recorded comorbidities.


Results
Serious operational problems (SOPs) were common; reported in 47 trusts (52.8%) and were present for 2,701 (of 21,716; 12.4%) trust days. Overall mortality was 37.7% (2,539 deaths). Admission during a period of SOPs was associated with a substantially increased mortality; adjusted odds ratio (OR) 1.34 (95% posterior credible interval (PCI): 1.07 to 1.68). Mortality was also associated with the proportion of a patient’s admission duration that had concurrent SOPs; OR 1.47 (95% PCI: 1.10 to 1.96) for mortality where SOPs were present for 100% compared to 0% of the stay.


Conclusion and relevance
Serious operational problems at the trust-level are associated with a significant increase in mortality in patients with COVID-19 admitted to critical care. The link isn’t necessarily causal, but this observation justifies further research to determine if a binary indicator might be a valid prognostic marker for deteriorating quality of care.

Abstract
. Background
. Type 2 diabetes (T2D) is common and increasing in prevalence. It is possible to prevent or delay T2D using lifestyle intervention programmes. Entry to these programmes is usually determined by a measure of glycaemia in the ‘intermediate’ range. This paper investigated the relationship between HbA1c and future diabetes risk and determined the impact of varying thresholds to identify those at high risk of developing T2D.
.
. Methods
. We studied 4227 participants without diabetes aged ≥ 40 years recruited to the Exeter 10,000 population cohort in South West England. HbA1c was measured at study recruitment with repeat HbA1c available as part of usual care. Absolute risk of developing diabetes within 5 years, defined by HbA1c ≥ 48 mmol/mol (6.5%), according to baseline HbA1c, was assessed by a flexible parametric survival model.
.
. Results
. The overall absolute 5-year risk (95% CI) of developing T2D in the cohort was 4.2% (3.6, 4.8%). This rose to 7.1% (6.1, 8.2%) in the 56% (n = 2358/4224) of participants classified ‘high-risk’ with HbA1c ≥ 39 mmol/mol (5.7%; ADA criteria). Under IEC criteria, HbA1c ≥ 42 mmol/mol (6.0%), 22% (n = 929/4277) of the cohort was classified high-risk with 5-year risk 14.9% (12.6, 17.2%). Those with the highest HbA1c values (44–47 mmol/mol [6.2–6.4%]) had much higher 5-year risk, 26.4% (22.0, 30.5%) compared with 2.1% (1.5, 2.6%) for 39–41 mmol/mol (5.7–5.9%) and 7.0% (5.4, 8.6%) for 42–43 mmol/mol (6.0–6.1%). Changing the entry criterion to prevention programmes from 39 to 42 mmol/mol (5.7–6.0%) reduced the proportion classified high-risk by 61%, and increased the positive predictive value (PPV) from 5.8 to 12.4% with negligible impact on the negative predictive value (NPV), 99.6% to 99.1%. Increasing the threshold further, to 44 mmol/mol (6.2%), reduced those classified high-risk by 59%, and markedly increased the PPV from 12.4 to 23.2% and had little impact on the NPV (99.1% to 98.5%).
.
. Conclusions
. A large proportion of people are identified as high-risk using current thresholds. Increasing the risk threshold markedly reduces the number of people that would be classified as high-risk and entered into prevention programmes, although this must be balanced against cases missed. Raising the entry threshold would allow limited intervention opportunities to be focused on those most likely to develop T2D.
.

AIMS/HYPOTHESIS: Among white European children developing type 1 diabetes, the otherwise common HLA haplotype DR15-DQ6 is rare, and highly protective. Adult-onset type 1 diabetes is now known to represent more overall cases than childhood onset, but it is not known whether DR15-DQ6 is protective in older-adult-onset type 1 diabetes. We sought to quantify DR15-DQ6 protection against type 1 diabetes as age of onset increased. METHODS: in two independent cohorts we assessed the proportion of type 1 diabetes cases presenting through the first 50 years of life with DR15-DQ6, compared with population controls. In the After Diabetes Diagnosis Research Support System-2 (ADDRESS-2) cohort (n = 1458) clinician-diagnosed type 1 diabetes was confirmed by positivity for one or more islet-specific autoantibodies. In UK Biobank (n = 2502), we estimated type 1 diabetes incidence rates relative to baseline HLA risk for each HLA group using Poisson regression. Analyses were restricted to white Europeans and were performed in three groups according to age at type 1 diabetes onset: 0-18 years, 19-30 years and 31-50 years. RESULTS: DR15-DQ6 was protective against type 1 diabetes through to age 50 years (OR 

One hundred years have passed since the discovery of insulin-an achievement that transformed diabetes from a fatal illness into a manageable chronic condition. The decades since that momentous achievement have brought ever more rapid innovation and advancement in diabetes research and clinical care. To celebrate the important work of the past century and help to chart a course for its continuation into the next, the Canadian Institutes of Health Research's Institute of Nutrition, Metabolism and Diabetes and the U.S. National Institutes of Health's National Institute of Diabetes and Digestive and Kidney Diseases recently held a joint international symposium, bringing together a cohort of researchers with diverse interests and backgrounds from both countries and beyond to discuss their collective quest to better understand the heterogeneity of diabetes and thus gain insights to inform new directions in diabetes treatment and prevention. This article summarizes the proceedings of that symposium, which spanned cutting-edge research into various aspects of islet biology, the heterogeneity of diabetic phenotypes, and the current state of and future prospects for precision medicine in diabetes.

ObjectiveIn this study, we describe the pattern of bed occupancy across England during the peak of the first wave of the COVID-19 pandemic.DesignDescriptive survey.SettingAll non-specialist secondary care providers in England from 27 March27to 5 June 2020.ParticipantsAcute (non-specialist) trusts with a type 1 (ie, 24 hours/day, consultant-led) accident and emergency department (n=125), Nightingale (field) hospitals (n=7) and independent sector secondary care providers (n=195).Main outcome measuresTwo thresholds for ‘safe occupancy’ were used: 85% as per the Royal College of Emergency Medicine and 92% as per NHS Improvement.ResultsAt peak availability, there were 2711 additional beds compatible with mechanical ventilation across England, reflecting a 53% increase in capacity, and occupancy never exceeded 62%. A consequence of the repurposing of beds meant that at the trough there were 8.7% (8508) fewer general and acute beds across England, but occupancy never exceeded 72%. The closest to full occupancy of general and acute bed (surge) capacity that any trust in England reached was 99.8%. For beds compatible with mechanical ventilation there were 326 trust-days (3.7%) spent above 85% of surge capacity and 154 trust-days (1.8%) spent above 92%. 23 trusts spent a cumulative 81 days at 100% saturation of their surge ventilator bed capacity (median number of days per trust=1, range: 1–17). However, only three sustainability and transformation partnerships (aggregates of geographically co-located trusts) reached 100% saturation of their mechanical ventilation beds.ConclusionsThroughout the first wave of the pandemic, an adequate supply of all bed types existed at a national level. However, due to an unequal distribution of bed utilisation, many trusts spent a significant period operating above ‘safe-occupancy’ thresholds despite substantial capacity in geographically co-located trusts, a key operational issue to address in preparing for future waves.

OBJECTIVES: to measure temporal trends in survival over time in people with severe coronavirus disease 2019 requiring critical care (high dependency unit or ICU) management, and to assess whether temporal variation in mortality was explained by changes in patient demographics and comorbidity burden over time. DESIGN: Retrospective observational cohort; based on data reported to the COVID-19 Hospitalisation in England Surveillance System. The primary outcome was in-hospital 30-day all-cause mortality. Unadjusted survival was estimated by calendar week of admission, and Cox proportional hazards models were used to estimate adjusted survival, controlling for age, sex, ethnicity, major comorbidities, and geographical region. SETTING: One hundred eight English critical care units. PATIENTS: all adult (18 yr +) coronavirus disease 2019 specific critical care admissions between March 1, 2020, and June 27, 2020. INTERVENTIONS: Not applicable. MEASUREMENTS AND MAIN RESULTS: Twenty-one thousand eighty-two critical care patients (high dependency unit n = 15,367; ICU n = 5,715) were included. Unadjusted survival at 30 days was lowest for people admitted in late March in both high dependency unit (71.6% survival) and ICU (58.0% survival). By the end of June, survival had improved to 92.7% in high dependency unit and 80.4% in ICU. Improvements in survival remained after adjustment for patient characteristics (age, sex, ethnicity, and major comorbidities) and geographical region. CONCLUSIONS: There has been a substantial improvement in survival amongst people admitted to critical care with coronavirus disease 2019 in England, with markedly higher survival rates in people admitted in May and June compared with those admitted in March and April. Our analysis suggests this improvement is not due to temporal changes in the age, sex, ethnicity, or major comorbidity burden of admitted patients.

BACKGROUND: the literature paints a complex picture of the association between mortality risk and ICU strain. In this study, we sought to determine if there is an association between mortality risk in intensive care units (ICU) and occupancy of beds compatible with mechanical ventilation, as a proxy for strain. METHODS: a national retrospective observational cohort study of 89 English hospital trusts (i.e. groups of hospitals functioning as single operational units). Seven thousand one hundred thirty-three adults admitted to an ICU in England between 2 April and 1 December, 2020 (inclusive), with presumed or confirmed COVID-19, for whom data was submitted to the national surveillance programme and met study inclusion criteria. A Bayesian hierarchical approach was used to model the association between hospital trust level (mechanical ventilation compatible), bed occupancy, and in-hospital all-cause mortality. Results were adjusted for unit characteristics (pre-pandemic size), individual patient-level demographic characteristics (age, sex, ethnicity, deprivation index, time-to-ICU admission), and recorded chronic comorbidities (obesity, diabetes, respiratory disease, liver disease, heart disease, hypertension, immunosuppression, neurological disease, renal disease). RESULTS: One hundred thirty-five thousand six hundred patient days were observed, with a mortality rate of 19.4 per 1000 patient days. Adjusting for patient-level factors, mortality was higher for admissions during periods of high occupancy (> 85% occupancy versus the baseline of 45 to 85%) [OR 1.23 (95% posterior credible interval (PCI): 1.08 to 1.39)]. In contrast, mortality was decreased for admissions during periods of low occupancy (

OBJECTIVES: to determine whether the previously described trend of improving mortality in people with coronavirus disease 2019 in critical care during the first wave was maintained, plateaued, or reversed during the second wave in United Kingdom, when B117 became the dominant strain. DESIGN: National retrospective cohort study. SETTING: all English hospital trusts (i.e. groups of hospitals functioning as single operational units), reporting critical care admissions (high dependency unit and ICU) to the Coronavirus Disease 2019 Hospitalization in England Surveillance System. PATIENTS: a total of 49,862 (34,336 high dependency unit and 15,526 ICU) patients admitted between March 1, 2020, and January 31, 2021 (inclusive). INTERVENTIONS: Not applicable. MEASUREMENTS AND MAIN RESULTS: the primary outcome was inhospital 28-day mortality by calendar month of admission, from March 2020 to January 2021. Unadjusted mortality was estimated, and Cox proportional hazard models were used to estimate adjusted mortality, controlling for age, sex, ethnicity, major comorbidities, social deprivation, geographic location, and operational strain (using bed occupancy as a proxy). Mortality fell to trough levels in June 2020 (ICU: 22.5% [95% CI, 18.2-27.4], high dependency unit: 8.0% [95% CI, 6.4-9.6]) but then subsequently increased up to January 2021: (ICU: 30.6% [95% CI, 29.0-32.2] and high dependency unit, 16.2% [95% CI, 15.3-17.1]). Comparing patients admitted during June-September 2020 with those admitted during December 2020-January 2021, the adjusted mortality was 59% (CI range, 39-82) higher in high dependency unit and 88% (CI range, 62-118) higher in ICU for the later period. This increased mortality was seen in all subgroups including those under 65. CONCLUSIONS: There was a marked deterioration in outcomes for patients admitted to critical care at the peak of the second wave of coronavirus disease 2019 in United Kingdom (December 2020-January 2021), compared with the post-first-wave period (June 2020-September 2020). The deterioration was independent of recorded patient characteristics and occupancy levels. Further research is required to determine to what extent this deterioration reflects the impact of the B117 variant of concern.

Abstract
Background
There is much interest in the use of prognostic and diagnostic prediction models in all areas of clinical medicine. The use of machine learning to improve prognostic and diagnostic accuracy in this area has been increasing at the expense of classic statistical models. Previous studies have compared performance between these two approaches but their findings are inconsistent and many have limitations. We aimed to compare the discrimination and calibration of seven models built using logistic regression and optimised machine learning algorithms in a clinical setting, where the number of potential predictors is often limited, and externally validate the models.

Methods
We trained models using logistic regression and six commonly used machine learning algorithms to predict if a patient diagnosed with diabetes has type 1 diabetes (versus type 2 diabetes). We used seven predictor variables (age, BMI, GADA islet-autoantibodies, sex, total cholesterol, HDL cholesterol and triglyceride) using a UK cohort of adult participants (aged 18–50 years) with clinically diagnosed diabetes recruited from primary and secondary care (n = 960, 14% with type 1 diabetes). Discrimination performance (ROC AUC), calibration and decision curve analysis of each approach was compared in a separate external validation dataset (n = 504, 21% with type 1 diabetes).

Results
Average performance obtained in internal validation was similar in all models (ROC AUC ≥ 0.94). In external validation, there were very modest reductions in discrimination with AUC ROC remaining ≥ 0.93 for all methods. Logistic regression had the numerically highest value in external validation (ROC AUC 0.95). Logistic regression had good performance in terms of calibration and decision curve analysis. Neural network and gradient boosting machine had the best calibration performance. Both logistic regression and support vector machine had good decision curve analysis for clinical useful threshold probabilities.

Conclusion
Logistic regression performed as well as optimised machine algorithms to classify patients with type 1 and type 2 diabetes. This study highlights the utility of comparing traditional regression modelling to machine learning, particularly when using a small number of well understood, strong predictor variables.

Despite the known heterogeneity of type 2 diabetes and variable response to glucose lowering medications, current evidence on optimal treatment is predominantly based on average effects in clinical trials rather than individual-level characteristics. A precision medicine approach based on treatment response would aim to improve on this by identifying predictors of differential drug response for people based on their characteristics and then using this information to select optimal treatment. Recent research has demonstrated robust and clinically relevant differential drug response with all noninsulin treatments after metformin (sulfonylureas, thiazolidinediones, dipeptidyl peptidase 4 [DPP-4] inhibitors, glucagon-like peptide 1 [GLP-1] receptor agonists, and sodium–glucose cotransporter 2 [SGLT2] inhibitors) using routinely available clinical features. This Perspective reviews this current evidence and discusses how differences in drug response could inform selection of optimal type 2 diabetes treatment in the near future. It presents a novel framework for developing and testing precision medicine–based strategies to optimize treatment, harnessing existing routine clinical and trial data sources. This framework was recently applied to demonstrate that “subtype” approaches, in which people are classified into subgroups based on features reflecting underlying pathophysiology, are likely to have less clinical utility compared with approaches that combine the same features as continuous measures in probabilistic “individualized prediction” models.

IntroductionTo identify risk factors, absolute risk, and impact on treatment discontinuation of genital infections with sodium-glucose co-transporter-2 inhibitors (SGLT2i).Research design and methodsWe assessed the relationship between baseline characteristics and genital infection in 21 004 people with type 2 diabetes initiating SGLT2i and 55 471 controls initiating dipeptidyl peptidase-4 inhibitors (DPP4i) in a UK primary care database. We assessed absolute risk of infection in those with key risk factors and the association between early genital infection and treatment discontinuation.ResultsGenital infection was substantially more common in those treated with SGLT2i (8.1% within 1 year) than DPP4i (1.8%). Key predictors of infection with SGLT2i were female sex (HR 3.64; 95% CI 3.23 to 4.11) and history of genital infection; <1 year before initiation (HR 4.38; 3.73 to 5.13), 1–5 years (HR 3.04; 2.64 to 3.51), and >5 years (HR 1.79; 1.55 to 2.07). Baseline HbA1c was not associated with infection risk for SGLT2i, in contrast to DPP4i where risk increased with higher HbA1c. One-year absolute risk of genital infection with SGLT2i was highest for those with a history of prior infection (females 23.7%, males 12.1%), compared with those without (females 10.8%, males 2.7%). Early genital infection was associated with a similar discontinuation risk for SGLT2i (HR 1.48; 1.21–1.80) and DPP4i (HR 1.58; 1.21–2.07).ConclusionsFemale sex and history of prior infection are simple features that can identify subgroups at greatly increased risk of genital infections with SGLT2i therapy. These data can be used to risk-stratify patients. High HbA1c is not a risk factor for genital infections with SGLT2i.

. OBJECTIVE
. To evaluate the association between metformin use and anemia risk in type 2 diabetes, and the time-course for this, in a randomized controlled trial (RCT) and real-world population data.
.
.
. RESEARCH DESIGN AND METHODS
. Anemia was defined as a hemoglobin measure of <11 g/dL. In the RCTs a Diabetes Outcome Progression Trial (ADOPT; n = 3,967) and UK Prospective Diabetes Study (UKPDS; n = 1,473), logistic regression was used to model anemia risk and nonlinear mixed models for change in hematological parameters. In the observational Genetics of Diabetes Audit and Research in Tayside Scotland (GoDARTS) population (n = 3,485), discrete-time failure analysis was used to model the effect of cumulative metformin exposure on anemia risk.
.
.
. RESULTS
. In ADOPT, compared with sulfonylureas, the odds ratio (OR) (95% CI) for anemia was 1.93 (1.10, 3.38) for metformin and 4.18 (2.50, 7.00) for thiazolidinediones. In UKPDS, compared with diet, the OR (95% CI) was 3.40 (1.98, 5.83) for metformin, 0.96 (0.57, 1.62) for sulfonylureas, and 1.08 (0.62, 1.87) for insulin. In ADOPT, hemoglobin and hematocrit dropped after metformin initiation by 6 months, with no further decrease after 3 years. In UKPDS, hemoglobin fell by 3 years in the metformin group compared with other treatments. At years 6 and 9, hemoglobin was reduced in all treatment groups, with no greater difference seen in the metformin group. In GoDARTS, each 1 g/day of metformin use was associated with a 2% higher annual risk of anemia.
.
.
. CONCLUSIONS
. Metformin use is associated with early risk of anemia in individuals with type 2 diabetes, a finding consistent across two RCTs and replicated in one real-world study. The mechanism for this early fall in hemoglobin is uncertain, but given the time course, is unlikely to be due to vitamin B12 deficiency alone.
.

. OBJECTIVE
. To describe the relationship between type 2 diabetes and all-cause mortality among adults with coronavirus disease 2019 (COVID-19) in the critical care setting.
.
.
. RESEARCH DESIGN AND METHODS
. This was a nationwide retrospective cohort study in people admitted to hospital in England with COVID-19 requiring admission to a high dependency unit (HDU) or intensive care unit (ICU) between 1 March 2020 and 27 July 2020. Cox proportional hazards models were used to estimate 30-day in-hospital all-cause mortality associated with type 2 diabetes, with adjustment for age, sex, ethnicity, obesity, and other major comorbidities (chronic respiratory disease, asthma, chronic heart disease, hypertension, immunosuppression, chronic neurological disease, chronic renal disease, and chronic liver disease).
.
.
. RESULTS
. A total of 19,256 COVID-19–related HDU and ICU admissions were included in the primary analysis, including 13,809 HDU (mean age 70 years) and 5,447 ICU (mean age 58 years) admissions. of those admitted, 3,524 (18.3%) had type 2 diabetes and 5,077 (26.4%) died during the study period. Patients with type 2 diabetes were at increased risk of death (adjusted hazard ratio [aHR] 1.23 [95% CI 1.14, 1.32]), and this result was consistent in HDU and ICU subsets. The relative mortality risk associated with type 2 diabetes decreased with higher age (age 18–49 years aHR 1.50 [95% CI 1.05, 2.15], age 50–64 years 1.29 [1.10, 1.51], and age ≥65 years 1.18 [1.09, 1.29]; P value for age–type 2 diabetes interaction = 0.002).
.
.
. CONCLUSIONS
. Type 2 diabetes may be an independent prognostic factor for survival in people with severe COVID-19 requiring critical care treatment, and in this setting the risk increase associated with type 2 diabetes is greatest in younger people.
.

AIMS: to measure the variation in prescribing of second-line non-insulin diabetes drugs. MATERIALS AND METHODS: We evaluated time trends for the period 1998 to 2016, using England's publicly available prescribing datasets, and stratified these by the order in which they were prescribed to patients using the Clinical Practice Research Datalink. We calculated the proportion of each class of diabetes drug as a percentage of the total per year. We evaluated geographical variation in prescribing using general practice-level data for the latest 12 months (to August 2017), with aggregation to Clinical Commissioning Groups. We calculated percentiles and ranges, and plotted maps. RESULTS: Prescribing of therapy after metformin is changing rapidly. Dipeptidyl peptidase-4 (DPP-4) inhibitor use has increased markedly, with DPP-4 inhibitors now the most common second-line drug (43% prescriptions in 2016). The use of sodium-glucose co-transporter-2 (SGLT-2) inhibitors also increased rapidly (14% new second-line, 27% new third-line prescriptions in 2016). There was wide geographical variation in choice of therapies and average spend per patient. In contrast, metformin was consistently used as a first-line treatment in accordance with guidelines. CONCLUSIONS: in England there is extensive geographical variation in the prescribing of diabetes drugs after metformin, and increasing use of higher-cost DPP-4 inhibitors and SGLT-2 inhibitors compared with low-cost sulphonylureas. Our findings strongly support the case for comparative effectiveness trials of current diabetes drugs.

OBJECTIVES: to describe the measurement of heart rate turbulence (HRT) after ventricular premature beats and compare HRT in healthy Doberman pinschers and those with dilated cardiomyopathy (DCM), with and without congestive heart failure (CHF). ANIMALS: Sixty-five client-owned Dobermans: 20 healthy (NORMAL), 31 with preclinical DCM and 14 with DCM and CHF (DCM + CHF). METHODS: a retrospective study of data retrieved from clinical records and ambulatory ECG (Holter) archives, including data collected previously for a large-scale prospective study of Dobermans with preclinical DCM. Holter data were reanalysed quantitatively, including conventional time-domain heart rate variability and the HRT parameters turbulence onset and turbulence slope. RESULTS: Heart rate turbulence could be measured in 58/65 dogs. Six Holter recordings had inadequate ventricular premature contractions (VPCs) and one exhibited VPCs too similar to sinus morphology. Heart rate turbulence parameter, turbulence onset, was significantly reduced in DCM dogs, whereas conventional heart rate variability measures were not. Heart rate variability and HRT markers were reduced in DCM + CHF dogs as expected. CONCLUSIONS: Heart rate turbulence can be measured from the majority of good quality standard canine 24-hour Holter recordings with >5 VPCs. Turbulence onset is significantly reduced in Dobermans with preclinical DCM which indicates vagal withdrawal early in the course of disease. Heart rate turbulence is a powerful prognostic indicator in human cardiac disease which can be measured from standard 24-hour ambulatory ECG (Holter) recordings using appropriate computer software. Further studies are warranted to assess whether HRT may be of prognostic value in dogs with preclinical DCM and in other canine cardiac disease.