Phenotype-based targeted treatment of SGLT2 inhibitors and GLP-1 receptor agonists in type 2 diabetes.

AIMS/HYPOTHESIS: A precision medicine approach in type 2 diabetes could enhance targeting specific glucose-lowering therapies to individual patients most likely to benefit. We aimed to use the recently developed Bayesian causal forest (BCF) method to develop and validate an individualised treatment selection algorithm for two major type 2 diabetes drug classes, sodium-glucose cotransporter 2 inhibitors (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP1-RA). METHODS: We designed a predictive algorithm using BCF to estimate individual-level conditional average treatment effects for 12-month glycaemic outcome (HbA1c) between SGLT2i and GLP1-RA, based on routine clinical features of 46,394 people with type 2 diabetes in primary care in England (Clinical Practice Research Datalink; 27,319 for model development, 19,075 for hold-out validation), with additional external validation in 2252 people with type 2 diabetes from Scotland (SCI-Diabetes [Tayside & Fife]). Differences in glycaemic outcome with GLP1-RA by sex seen in clinical data were replicated in clinical trial data (HARMONY programme: liraglutide [n=389] and albiglutide [n=1682]). As secondary outcomes, we evaluated the impacts of targeting therapy based on glycaemic response on weight change, tolerability and longer-term risk of new-onset microvascular complications, macrovascular complications and adverse kidney events. RESULTS: Model development identified marked heterogeneity in glycaemic response, with 4787 (17.5%) of the development cohort having a predicted HbA1c benefit >3 mmol/mol (>0.3%) with SGLT2i over GLP1-RA and 5551 (20.3%) having a predicted HbA1c benefit >3 mmol/mol with GLP1-RA over SGLT2i. Calibration was good in hold-back validation, and external validation in an independent Scottish dataset identified clear differences in glycaemic outcomes between those predicted to benefit from each therapy. Sex, with women markedly more responsive to GLP1-RA, was identified as a major treatment effect modifier in both the UK observational datasets and in clinical trial data: HARMONY-7 liraglutide (GLP1-RA): 4.4 mmol/mol (95% credible interval [95% CrI] 2.2, 6.3) (0.4% [95% CrI 0.2, 0.6]) greater response in women than men. Targeting the two therapies based on predicted glycaemic response was also associated with improvements in short-term tolerability and long-term risk of new-onset microvascular complications. CONCLUSIONS/INTERPRETATION: Precision medicine approaches can facilitate effective individualised treatment choice between SGLT2i and GLP1-RA therapies, and the use of routinely collected clinical features for treatment selection could support low-cost deployment in many countries.

Comparison of causal forest and regression-based approaches to evaluate treatment effect heterogeneity: an application for type 2 diabetes precision medicine.

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.

Azathioprine with Allopurinol Is a Promising First-Line Therapy for Inflammatory Bowel Diseases.

BACKGROUND: Beneficial response to first-line immunosuppressive azathioprine in patients with inflammatory bowel disease (IBD) is low due to high rates of adverse events. Co-administrating allopurinol has been shown to improve tolerability. However, data on this co-therapy as first-line treatment are scarce. AIM: Retrospective comparison of long-term effectiveness and safety of first-line low-dose azathioprine-allopurinol co-therapy (LDAA) with first-line azathioprine monotherapy (AZAm) in patients with IBD without metabolite monitoring. METHODS: Clinical benefit was defined as ongoing therapy without initiation of steroids, biologics or surgery. Secondary outcomes included CRP, HBI/SCCAI, steroid withdrawal and adverse events. RESULTS: In total, 166 LDAA and 118 AZAm patients (median follow-up 25 and 27 months) were evaluated. Clinical benefit was more frequently observed in LDAA patients at 6 months (74% vs. 53%, p = 0.0003), 12 months (54% vs. 37%, p = 0.01) and in the long-term (median 36 months; 37% vs. 24%, p = 0.04). Throughout follow-up, AZAm patients were 60% more likely to fail therapy, due to a higher intolerance rate (45% vs. 26%, p = 0.001). Only 73% of the effective AZA dose was tolerated in AZAm patients, while LDAA could be initiated and maintained at its target dose. Incidence of myelotoxicity and elevated liver enzymes was similar in both cohorts, and both conditions led to LDAA withdrawal in only 2%. Increasing allopurinol from 100 to 200-300 mg/day significantly lowered liver enzymes in 5/6 LDAA patients with hepatotoxicity. CONCLUSIONS: Our poor AZAm outcomes emphasize that optimization of azathioprine is needed. We demonstrated a long-term safe and more effective profile of first-line LDAA. This co-therapy may therefore be considered standard first-line immunosuppressive.

The association between mechanical ventilator compatible bed occupancy and mortality risk in intensive care patients with COVID-19: a national retrospective cohort study.

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 (< 45% relative to the baseline) [OR 0.83 (95% PCI 0.75 to 0.94)]. CONCLUSION: Increasing occupancy of beds compatible with mechanical ventilation, a proxy for operational strain, is associated with a higher mortality risk for individuals admitted to ICU. Further research is required to establish if this is a causal relationship or whether it reflects strain on other operational factors such as staff. If causal, the result highlights the importance of strategies to keep ICU occupancy low to mitigate the impact of this type of resource saturation.

Improving Survival of Critical Care Patients With Coronavirus Disease 2019 in England: A National Cohort Study, March to June 2020.

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.

Trends in 28-Day Mortality of Critical Care Patients With Coronavirus Disease 2019 in the United Kingdom: A National Cohort Study, March 2020 to January 2021.

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.

The Trail Making test: a study of its ability to predict falls in the acute neurological in-patient population.

OBJECTIVE: To determine whether tests of cognitive function and patient-reported outcome measures of motor function can be used to create a machine learning-based predictive tool for falls. DESIGN: Prospective cohort study. SETTING: Tertiary neurological and neurosurgical center. SUBJECTS: In all, 337 in-patients receiving neurosurgical, neurological, or neurorehabilitation-based care. MAIN MEASURES: Binary (Y/N) for falling during the in-patient episode, the Trail Making Test (a measure of attention and executive function) and the Walk-12 (a patient-reported measure of physical function). RESULTS: The principal outcome was a fall during the in-patient stay ( n = 54). The Trail test was identified as the best predictor of falls. Moreover, addition of other variables, did not improve the prediction (Wilcoxon signed-rank P < 0.001). Classical linear statistical modeling methods were then compared with more recent machine learning based strategies, for example, random forests, neural networks, support vector machines. The random forest was the best modeling strategy when utilizing just the Trail Making Test data (Wilcoxon signed-rank P < 0.001) with 68% (± 7.7) sensitivity, and 90% (± 2.3) specificity. CONCLUSION: This study identifies a simple yet powerful machine learning (Random Forest) based predictive model for an in-patient neurological population, utilizing a single neuropsychological test of cognitive function, the Trail Making test.

Systematic Review of Health-Related Work Outcome Measures and Quality Criteria-Based Evaluations of Their Psychometric Properties.

OBJECTIVE: To examine the state of psychometric validation in the health-related work outcome literature. DATA SOURCES: We searched PubMed, PubMed Central, CINAHL, Embase (plus Embase Classic), and PsycINFO from inception to January 2016 using the following search terms: stroke, multiple sclerosis, epilepsy, spinal cord injury, brain injury, musculoskeletal disease, work, absenteeism, presenteeism, occupation, employment, job, outcome measure, assessment, work capacity evaluation, scale, and questionnaire. STUDY SELECTION: From the 22,676 retrieved abstracts, 597 outcome measures were identified. Inclusion was based on content analysis. There were 95 health-related work outcome measures retained; of these, 2 were treated as outliers and therefore are discussed separately. All 6 authors individually organized the 93 remaining scales based on their content. DATA EXTRACTION: A follow-up search using the same sources, and time period, with the name of the outcome measures and the terms psychometric, reliability, validity, and responsiveness, identified 263 unique classical test theory psychometric property datasets for the 93 tools. An assessment criterion for psychometric properties was applied to each article, and where consensus was not achieved, the rating delivered by most of the assessors was reported. DATA SYNTHESIS: Of the articles reported, 18 reporting psychometric data were not accessible and therefore could not be assessed. There were 39 that scored <20% of the maximum achievable score, 106 scored between 20% and 40%, 82 scored between 40% and 60%, 15 scored between 60% and 80%, and only 1 scored >80%. The 3 outcome measures associated with the highest scoring datasets were the Sheehan Disability Scale, the Fear Avoidance Beliefs Questionnaire, and the assessment of the Subjective Handicap of Epilepsy. Finally, only 2 psychometric validation datasets reported the complete set of baseline psychometric properties. CONCLUSIONS: This systematic review highlights the current limitations of the health-related work outcome measure literature, including the limited number of robust tools available.

Risk factor associations for severe COVID-19, influenza and pneumonia in people with diabetes to inform future pandemic preparations: UK population-based cohort study.

OBJECTIVE: This study aimed to compare clinical and sociodemographic risk factors for severe COVID-19, influenza and pneumonia, in people with diabetes. DESIGN: Population-based cohort study. SETTING: UK primary care records (Clinical Practice Research Datalink) linked to mortality and hospital records. PARTICIPANTS: Individuals with type 1 and type 2 diabetes (COVID-19 cohort: n=43 033 type 1 diabetes and n=584 854 type 2 diabetes, influenza and pneumonia cohort: n=42 488 type 1 diabetes and n=585 289 type 2 diabetes). PRIMARY AND SECONDARY OUTCOME MEASURES: COVID-19 hospitalisation from 1 February 2020 to 31 October 2020 (pre-COVID-19 vaccination roll-out), and influenza and pneumonia hospitalisation from 1 September 2016 to 31 May 2019 (pre-COVID-19 pandemic). Secondary outcomes were COVID-19 and pneumonia mortality. Associations between clinical and sociodemographic risk factors and each outcome were assessed using multivariable Cox proportional hazards models. In people with type 2 diabetes, we explored modifying effects of glycated haemoglobin (HbA1c) and body mass index (BMI) by age, sex and ethnicity. RESULTS: In type 2 diabetes, poor glycaemic control and severe obesity were consistently associated with increased risk of hospitalisation for COVID-19, influenza and pneumonia. The highest HbA1c and BMI-associated relative risks were observed in people aged under 70 years. Sociodemographic-associated risk differed markedly by respiratory infection, particularly for ethnicity. Compared with people of white ethnicity, black and south Asian groups had a greater risk of COVID-19 hospitalisation, but a lesser risk of pneumonia hospitalisation. Risk factor associations for type 1 diabetes and for type 2 diabetes mortality were broadly consistent with the primary analysis. CONCLUSIONS: Clinical risk factors of high HbA1c and severe obesity are consistently associated with severe outcomes from COVID-19, influenza and pneumonia, especially in younger people. In contrast, associations with sociodemographic risk factors differed by type of respiratory infection. This emphasises that risk stratification should be specific to individual respiratory infections.

A Methodological Framework for Assessing the Benefit of SARS-CoV-2 Vaccination following Previous Infection: Case Study of Five- to Eleven-Year-Olds.

Vaccination rates against SARS-CoV-2 in children aged five to eleven years remain low in many countries. The current benefit of vaccination in this age group has been questioned given that the large majority of children have now experienced at least one SARS-CoV-2 infection. However, protection from infection, vaccination or both wanes over time. National decisions on offering vaccines to this age group have tended to be made without considering time since infection. There is an urgent need to evaluate the additional benefits of vaccination in previously infected children and under what circumstances those benefits accrue. We present a novel methodological framework for estimating the potential benefits of COVID-19 vaccination in previously infected children aged five to eleven, accounting for waning. We apply this framework to the UK context and for two adverse outcomes: hospitalisation related to SARS-CoV-2 infection and Long Covid. We show that the most important drivers of benefit are: the degree of protection provided by previous infection; the protection provided by vaccination; the time since previous infection; and future attack rates. Vaccination can be very beneficial for previously infected children if future attack rates are high and several months have elapsed since the previous major wave in this group. Benefits are generally larger for Long Covid than hospitalisation, because Long Covid is both more common than hospitalisation and previous infection offers less protection against it. Our framework provides a structure for policy makers to explore the additional benefit of vaccination across a range of adverse outcomes and different parameter assumptions. It can be easily updated as new evidence emerges.

Where next for partial randomisation of research funding? The feasibility of RCTs and alternatives.

We outline essential considerations for any study of partial randomisation of research funding, and consider scenarios in which randomised controlled trials (RCTs) would be feasible and appropriate. We highlight the interdependence of target outcomes, sample availability and statistical power for determining the cost and feasibility of a trial. For many choices of target outcome, RCTs may be less practical and more expensive than they at first appear (in large part due to issues pertaining to sample size and statistical power). As such, we briefly discuss alternatives to RCTs. It is worth noting that many of the considerations relevant to experiments on partial randomisation may also apply to other potential experiments on funding processes (as described in The Experimental Research Funder's Handbook. RoRI, June 2022).

Qualitative interviews to understand methods and systems used to collect ethnicity information in health administrative data sources in England.

Background: This article is one of a series aiming to inform analytical methods to improve comparability of estimates of ethnic health disparities based on different sources. This article explores the quality of ethnicity data and identifies potential sources of bias when ethnicity information is collected in three key NHS data sources. Future research can build on these findings to explore analytical methods to mitigate biases. Methods: Thematic analysis of semi-structured qualitative interviews to explore potential sources of error and bias in the process of collecting ethnicity information across three NHS data sources: General Practice Extraction Service (GPES) Data for Pandemic Planning and Research (GDPPR), Hospital Episode Statistics (HES) and Improving Access to Psychological Therapies (IAPT). The study included feedback from 22 experts working on different aspects of health admin data collection for England (including staff from NHS Digital, IT system suppliers and relevant healthcare service providers). Results: Potential sources of error and bias were identified across data collection, data processing and quality assurance processes. Similar issues were identified for all three sources. Our analysis revealed three main themes which can result in bias and inaccuracies in ethnicity data recorded: data infrastructure challenges, human challenges, and institutional challenges. Conclusions: Findings highlighted that analysts using health admin data should be aware of the main sources of potential error and bias in health admin data, and be mindful that the main sources of error identified are more likely to affect the ethnicity data for ethnic minority groups. Where possible, analysts should describe and seek to account for this bias in their research.

The current landscape of software tools for the climate-sensitive infectious disease modelling community.

Climate-sensitive infectious disease modelling is crucial for public health planning and is underpinned by a complex network of software tools. We identified only 37 tools that incorporated both climate inputs and epidemiological information to produce an output of disease risk in one package, were transparently described and validated, were named (for future searching and versioning), and were accessible (ie, the code was published during the past 10 years or was available on a repository, web platform, or other user interface). We noted disproportionate representation of developers based at North American and European institutions. Most tools (n=30 [81%]) focused on vector-borne diseases, and more than half (n=16 [53%]) of these tools focused on malaria. Few tools (n=4 [11%]) focused on food-borne, respiratory, or water-borne diseases. The under-representation of tools for estimating outbreaks of directly transmitted diseases represents a major knowledge gap. Just over half (n=20 [54%]) of the tools assessed were described as operationalised, with many freely available online.

Hospital admissions linked to SARS-CoV-2 infection in children and adolescents: cohort study of 3.2 million first ascertained infections in England.

OBJECTIVE: To describe hospital admissions associated with SARS-CoV-2 infection in children and adolescents. DESIGN: Cohort study of 3.2 million first ascertained SARS-CoV-2 infections using electronic health care record data. SETTING: England, July 2020 to February 2022. PARTICIPANTS: About 12 million children and adolescents (age <18 years) who were resident in England. MAIN OUTCOME MEASURES: Ascertainment of a first SARS-CoV-2 associated hospital admissions: due to SARS-CoV-2, with SARS-CoV-2 as a contributory factor, incidental to SARS-CoV-2 infection, and hospital acquired SARS-CoV-2. RESULTS: 3 226 535 children and adolescents had a recorded first SARS-CoV-2 infection during the observation period, and 29 230 (0.9%) infections involved a SARS-CoV-2 associated hospital admission. The median length of stay was 2 (interquartile range 1-4) days) and 1710 of 29 230 (5.9%) SARS-CoV-2 associated admissions involved paediatric critical care. 70 deaths occurred in which covid-19 or paediatric inflammatory multisystem syndrome was listed as a cause, of which 55 (78.6%) were in participants with a SARS-CoV-2 associated hospital admission. SARS-CoV-2 was the cause or a contributory factor in 21 000 of 29 230 (71.8%) participants who were admitted to hospital and only 380 (1.3%) participants acquired infection as an inpatient and 7855 (26.9%) participants were admitted with incidental SARS-CoV-2 infection. Boys, younger children (<5 years), and those from ethnic minority groups or areas of high deprivation were more likely to be admitted to hospital (all P<0.001). The covid-19 vaccination programme in England has identified certain conditions as representing a higher risk of admission to hospital with SARS-CoV-2: 11 085 (37.9%) of participants admitted to hospital had evidence of such a condition, and a further 4765 (16.3%) of participants admitted to hospital had a medical or developmental health condition not included in the vaccination programme's list. CONCLUSIONS: Most SARS-CoV-2 associated hospital admissions in children and adolescents in England were due to SARS-CoV-2 or SARS-CoV-2 was a contributory factor. These results should inform future public health initiatives and research.

The value of standards for health datasets in artificial intelligence-based applications.

Artificial intelligence as a medical device is increasingly being applied to healthcare for diagnosis, risk stratification and resource allocation. However, a growing body of evidence has highlighted the risk of algorithmic bias, which may perpetuate existing health inequity. This problem arises in part because of systemic inequalities in dataset curation, unequal opportunity to participate in research and inequalities of access. This study aims to explore existing standards, frameworks and best practices for ensuring adequate data diversity in health datasets. Exploring the body of existing literature and expert views is an important step towards the development of consensus-based guidelines. The study comprises two parts: a systematic review of existing standards, frameworks and best practices for healthcare datasets; and a survey and thematic analysis of stakeholder views of bias, health equity and best practices for artificial intelligence as a medical device. We found that the need for dataset diversity was well described in literature, and experts generally favored the development of a robust set of guidelines, but there were mixed views about how these could be implemented practically. The outputs of this study will be used to inform the development of standards for transparency of data diversity in health datasets (the STANDING Together initiative).

Psychometric analysis of the Glasgow Coma Scale and its sub-scale scores in a national retrospective cohort of patients with traumatic injuries.

OBJECTIVES: To determine the psychometric validity, using Rasch analysis, of summing the three constituent parts of the Glasgow Coma Scale (GCS). DESIGN: National (registry-based) retrospective study. SETTING: England and Wales. PATIENTS: All individuals who sustained a traumatic injury and were: admitted for more than three days; required critical care resources; transferred for specialist management; or who died from their injuries. MAIN OUTCOMES AND MEASURES: Demographic information (i.e., age at time of injury, and sex), item sub-scores of the first available GCS (either completed by the attending paramedics or on arrival to hospital), injury severity as denoted by the Injury Severity Scale (ISS), and outcome (survival to hospital discharge or 30-days post-injury, whichever is earliest). RESULTS: 321,203 cases between 2008 and 2017. 55.9% were male, the median age was 62.7 years (IQR 44.2-80.8), the median ISS was 9 (IQR 9 to 17), and 6.6% were deceased at 30 days. The reliability statistics suggest that when the extreme scores (i.e. 3 and 15) are accounted for, that there is only sufficient consistency to support the separation of injuries into 3 broad categories, e.g. mild, moderate and severe. As extreme scores don't impact Rasch item calibrations, subsequent analysis was restricted to the 48,417 non-extreme unique cases. Overall fit to the Rasch model was poor across all analyses (p < 0.0001). Through a combination of empirical evidence and clinical reasoning, item response categories were collapsed to provide a post-hoc scoring amendment. Whilst the modifications improved the function of the individual items, there is little evidence to support them meaningfully contributing to a total score that can be interpreted on an interval scale. CONCLUSION AND RELEVANCE: The GCS does not perform in a psychometrically robust manner in a national retrospective cohort of individuals who have experienced a traumatic injury, even after post-hoc correction.

Development of a treatment selection algorithm for SGLT2 and DPP-4 inhibitor therapies in people with type 2 diabetes: a retrospective cohort study.

BACKGROUND: Current treatment guidelines do not provide recommendations to support the selection of treatment for most people with type 2 diabetes. We aimed to develop and validate an algorithm to allow selection of optimal treatment based on glycaemic response, weight change, and tolerability outcomes when choosing between SGLT2 inhibitor or DPP-4 inhibitor therapies. METHODS: In this retrospective cohort study, we identified patients initiating SGLT2 and DPP-4 inhibitor therapies after Jan 1, 2013, from the UK Clinical Practice Research Datalink (CPRD). We excluded those who received SGLT2 or DPP-4 inhibitors as first-line treatment or insulin at the same time, had estimated glomerular filtration rate (eGFR) of less than 45 mL/min per 1·73 m2, or did not have a valid baseline glycated haemoglobin (HbA1c) measure (<53 or ≥120 mmol/mol). The primary efficacy outcome was the HbA1c value reached 6 months after drug initiation, adjusted for baseline HbA1c. Clinical features associated with differential HbA1c outcome on the two therapies were identified in CPRD (n=26 877), and replicated in reanalysis of 14 clinical trials (n=10 414). An algorithm to predict individual-level differential HbA1c outcome on the two therapies was developed in CPRD (derivation; n=14 069) and validated in head-to-head trials (n=2499) and CPRD (independent validation; n=9376). In CPRD, we further explored heterogeneity in 6-month weight change and treatment discontinuation. FINDINGS: Among 10 253 patients initiating SGLT2 inhibitors and 16 624 patients initiating DPP-4 inhibitors in CPRD, baseline HbA1c, age, BMI, eGFR, and alanine aminotransferase were associated with differential HbA1c outcome with SGLT2 inhibitor and DPP-4 inhibitor therapies. The median age of participants was 62·0 years (IQR 55·0-70·0). 10 016 (37·3%) were women and 16 861 (62·7%) were men. An algorithm based on these five features identified a subgroup, representing around four in ten CPRD patients, with a 5 mmol/mol or greater observed benefit with SGLT2 inhibitors in all validation cohorts (CPRD 8·8 mmol/mol [95% CI 7·8-9·8]; CANTATA-D and CANTATA-D2 trials 5·8 mmol/mol [3·9-7·7]; BI1245.20 trial 6·6 mmol/mol [2·2-11·0]). In CPRD, predicted differential HbA1c response with SGLT2 inhibitor and DPP-4 inhibitor therapies was not associated with weight change. Overall treatment discontinuation within 6 months was similar in patients predicted to have an HbA1c benefit with SGLT2 inhibitors over DPP-4 inhibitors (median 15·2% [13·2-20·3] vs 14·4% [12·9-16·7]). A smaller subgroup predicted to have greater HbA1c reduction with DPP-4 inhibitors were twice as likely to discontinue SGLT2 inhibitors than DPP-4 inhibitors (median 26·8% [23·4-31·0] vs 14·8% [12·9-16·8]). INTERPRETATION: A validated treatment selection algorithm for SGLT2 inhibitor and DPP-4 inhibitor therapies can support decisions on optimal treatment for people with type 2 diabetes. FUNDING: BHF-Turing Cardiovascular Data Science Award and the UK Medical Research Council.

Reporting guideline for the early-stage clinical evaluation of decision support systems driven by artificial intelligence: DECIDE-AI.

A growing number of artificial intelligence (AI)-based clinical decision support systems are showing promising performance in preclinical, in silico evaluation, but few have yet demonstrated real benefit to patient care. Early-stage clinical evaluation is important to assess an AI system's actual clinical performance at small scale, ensure its safety, evaluate the human factors surrounding its use and pave the way to further large-scale trials. However, the reporting of these early studies remains inadequate. The present statement provides a multi-stakeholder, consensus-based reporting guideline for the Developmental and Exploratory Clinical Investigations of DEcision support systems driven by Artificial Intelligence (DECIDE-AI). We conducted a two-round, modified Delphi process to collect and analyze expert opinion on the reporting of early clinical evaluation of AI systems. Experts were recruited from 20 pre-defined stakeholder categories. The final composition and wording of the guideline was determined at a virtual consensus meeting. The checklist and the Explanation & Elaboration (E&E) sections were refined based on feedback from a qualitative evaluation process. In total, 123 experts participated in the first round of Delphi, 138 in the second round, 16 in the consensus meeting and 16 in the qualitative evaluation. The DECIDE-AI reporting guideline comprises 17 AI-specific reporting items (made of 28 subitems) and ten generic reporting items, with an E&E paragraph provided for each. Through consultation and consensus with a range of stakeholders, we developed a guideline comprising key items that should be reported in early-stage clinical studies of AI-based decision support systems in healthcare. By providing an actionable checklist of minimal reporting items, the DECIDE-AI guideline will facilitate the appraisal of these studies and replicability of their findings.

COVID-19 trajectories among 57 million adults in England: a cohort study using electronic health records.

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.