Search results
Found 5552 matches for
The home of collaborative applied health and social care research in Oxford and the Thames Valley.
Model development for bespoke large language models for digital triage assistance in mental health care.
Contemporary large language models (LLMs) may have utility for processing unstructured, narrative free-text clinical data contained in electronic health records (EHRs) - a particularly important use-case for mental health where a majority of routinely-collected patient data lacks structured, machine-readable content. A significant problem for the United Kingdom's National Health Service (NHS) are the long waiting lists for specialist mental healthcare. According to NHS data (NHS Digital, 2024), in each month of 2023, there were between 370,000 and 470,000 individual new referrals into secondary mental healthcare services. Referrals must be triaged by clinicians, using clinical information contained in the patient's EHR to arrive at a decision about the most appropriate mental healthcare team to assess and potentially treat these patients. The ability to efficiently recommend a relevant team by ingesting potentially voluminous clinical notes could help services both reduce referral waiting times and with the right technology, improve the evidence available to justify triage decisions. We present and evaluate three different approaches for LLM-based, end-to-end ingestion of variable-length clinical EHR data to assist clinicians when triaging referrals. Our model is able to deliver triage recommendations consistent with existing clinical practices and its architecture was implemented on a single GPU, making it practical for implementation in resource-limited NHS environments where private implementations of LLM technology will be necessary to ensure confidential clinical data are appropriately controlled and governed. Code available at: https://github.com/NtaylorOX/BespokeLLM_Triage.
Moderators of the Effects of a Digital Parenting Intervention on Child Conduct and Emotional Problems Implemented During the COVID-19 Pandemic: Results From a Secondary Analysis of Data From the Supporting Parents and Kids Through Lockdown Experiences (SPARKLE) Randomized Controlled Trial.
BACKGROUND: A smartphone app, Parent Positive, was developed to help parents manage their children's conduct and emotional problems during the COVID-19 pandemic. A randomized controlled trial, Supporting Parents and Kids Through Lockdown Experiences (SPARKLE), found Parent Positive to be effective in reducing children's emotional problems. However, app effectiveness may be influenced by a range of child, family, socioeconomic, and pandemic-related factors. OBJECTIVE: This study examined whether baseline factors related to the child, family, and socioeconomic status, as well as pandemic-related disruption circumstances, moderated Parent Positive's effects on child conduct and emotional problems at 1- and 2-month follow-up. METHODS: This study was a secondary exploratory analysis of SPARKLE data. The data set included 646 children (4-10 years of age) with parents randomized to either Parent Positive (n=320) or follow-up as usual (n=326). Candidate baseline moderators included child age, gender, attention-deficit/hyperactivity disorder symptoms, parental psychological distress, family conflict, household income, employment status, household overcrowding, and pandemic-related disruption risk (ie, homeschooling, lockdown status, and isolation status). Child conduct and emotional problem outcomes measured at 1- (T2) and 2-months (T3) post randomization were analyzed using linear mixed-effects analysis of covariance models adjusting for baseline (T1) measure of outcome and including intervention and intervention by time point interaction terms allowing for different effects at the 2 time points. Moderation of intervention effects by baseline factors was assessed by replacing the intervention by time interaction terms with intervention by time point by baseline moderator interaction terms. RESULTS: Child gender was a significant moderator of the Parent Positive versus follow-up as usual effect on emotional problems (B=0.72, 95% CI 0.12-1.33; P=.02). Specifically, the effect of Parent Positive was close to significant (T2: B=-0.41, 95% CI -0.82 to 0.0004; P=.05) or significant (T3: B=-0.76, 95% CI -1.22 to -0.30; P
Validation of the parent-report pandemic anxiety scale (PAS-P) in the context of COVID-19
To be able to develop effective policy and targeted support for children and young people, it is vital to develop and validate measures that enable us to understand what aspects of pandemics are associated with anxiety and stress across a wide age range. We examined the psychometric properties of the Pandemic Anxiety Scale– Parent-report (PAS-P), which measures levels of child and adolescent pandemic-related anxieties. Factor structure, reliability, and convergent and discriminant validity of the PAS-P was assessed in a convenience sample of parents/carers (N = 8410) over at three time points in 2020 when COVID-19 case rates and restrictions varied. Factor structure was identified via two exploratory factor analyses (EFAs; n = 5601 and n = 1005) and then tested using confirmatory factor analysis (CFA; n = 800), measurement invariance tests, and a longitudinal CFA (n = 1651). Two factors structure for disease and consequence anxiety were observed across both EFAs and were found to have good fit in the CFAs. The PAS-P demonstrated good internal consistency and discriminant validity, as well as partial scalar invariance in latent construct measurement across child age, gender, and time. The PAS-P is a robust parent-report measure of two distinct forms of pandemic-related anxiety, suitable for reporting on children and adolescents aged 4-16 years. Although the scale has been validated in the context of the COVID-19 pandemic, it is not limited to this specific pandemic and, with minor wording modifications, may be a reliable tool in other health epidemic contexts.
Evaluating video and hybrid group consultations in general practice: mixed-methods, participatory study protocol (TOGETHER 2).
BACKGROUND: General practice is facing an unprecedented challenge in managing the consequences of the pandemic. In the midst of a policy drive to balance remote and in-person service provision, substantial workload pressures remain, together with increasing prevalence of long-term conditions, and declining staff numbers and morale. To address these challenges, some practices in the UK have been delivering video and hybrid group consultations (VHGCs) providing clinical care to multiple patients at the same time. Despite positive initial findings and enthusiasm, there are still gaps in our understanding of the influence VHGCs have on patient experience, healthcare utilisation, quality, safety, equity and affordability. OBJECTIVES: To generate an in-depth understanding of VHGCs for chronic conditions in general practice, surface assumptions and sociotechnical dynamics, inform practice and extend theorisation. METHODS: Mixed-methods, multi-site research study using co-design and participatory methods, from qualitative, quantitative and cost-related perspectives. WP1 includes a national, cross-sectional survey on VHGC provision across the UK. In WP2 we will engage patients and general practice staff in co-design workshops to develop VHGC models with emphasis on digital inclusion and equity. In WP3 we will carry out a mixed-methods process evaluation in up to 10 GP practices across England (5 sites already running VHGCs and 5 comparison sites). Qualitative methods will include interviews, focus groups and ethnographic observation to examine the experiences of patients, carers, clinical and non-clinical NHS staff, commissioners and policy-makers. Quantitative methods will examine the impact of VHGCs on healthcare utilisation in primary and secondary care, patient satisfaction, engagement and activation. We will also assess value for money of group and individual care models from a health economics perspective. CONCLUSIONS: We aim to develop transferable learning on sociotechnical change in healthcare delivery, using VHGCs as an exemplar of technology-supported innovation. Findings will also inform the design of a future study.
Development of a health state classification system for the PedsQL™ 4.0 Generic Core Scales for preference-based valuation in Australia.
OBJECTIVES: Pediatric Quality of Life InventoryTM Version 4.0 Generic Core Scales (PedsQL GCS), comprising 23 items covering four sub-scales (physical, emotional, social, and school functioning), is a widely applied generic measure of childhood health-related quality of life, but does not provide health utilities for cost-effectiveness-based decision-making. This study aimed to develop a reduced item version of PedsQL GCS amenable to health utility derivation in Australia. METHODS: Data sources were two cohorts of the Longitudinal Study of Australian Children, including proxy responses for all PedsQL GCS versions (Toddlers, Young Children, Children, Teens), and the CheckPoint sample containing child self-report to the Children version. Three analytic samples were: CheckPoint sample (n=1,874); Mallinson sample containing one measurement per child from one of the Young Children, Children or Teens versions (n=7,855); and Toddlers sample (n=7,401). Exploratory and confirmatory factor analyses assessed dimensionality. Psychometric analyses used Rasch and classical criteria on three randomly selected subsamples (n=500) per sample. Item selection prioritised psychometric performance in CheckPoint sample, also considering performance in other samples and conceptual content. RESULTS: Dimensionality assessments did not generate an alternative empirical structure for the measure, and psychometric analyses were conducted on the original four sub-scales. The selected items were: 'Get aches and pains' for physical functioning; 'Feel sad/blue' for emotional functioning; 'Other kids not friends' for social functioning; and 'Keeping up with school work' for school functioning. CONCLUSIONS: The final four-item set, pending further psychometric validation and valuation, can generate health utilities from the widely used PedsQL GCS to inform cost-effectiveness-based decision-making.
Student perspectives on school-based social workers: a mixed-methods study
Purpose: School social work, in various forms, is well established internationally and has a growing evidence base. Yet existing research focuses on professional perspectives rather than those of students. This study aims to fill this gap by exploring secondary school student perspectives of having social workers in schools (SWIS). It was undertaken in England as part of the SWIS trial, which tested whether secondary school-based social workers could improve child safety and well-being, identify issues more quickly and reduce the need for statutory services. Design/methodology/approach: A mixed-methods approach comprised a survey (n = 1,998 students) and interviews (n = 27 students). Surveys included questions on awareness, understanding, interaction with and attitudes towards the school-based social workers. Interviews involved a Q-sort activity followed by semi-structured questions on general attitudes and experiences. The Q-sort characterised prominent perspectives and how many students subscribed to them. Findings: Students were broadly positive about having a social worker in their school in the survey and interviews. Two prominent perspectives on SWIS were identified. The first (n = 17) was defined by students feeling positively overall and strongly agreeing that they trusted the social worker. The second (n = 4) was mixed in sentiment, defined by some anxiety about working with the social worker. In interviews, students relayed that social workers were easily accessible, offered emotional support and acted as a bridge between school and home. Originality/value: To the best of the authors’ knowledge, this study is the first to quantify student perspectives on having social workers at school and evidence attitudes and experiences about school-based social work as practiced during the SWIS trial.
Lessons to be learned from test evaluations during the COVID-19 pandemic: RSS Working Group’s Report on Diagnostic Tests
The coronavirus disease (Covid-19) pandemic raised challenges for everyday life. Development of new diagnostic tests was necessary, but under such enormous pressure risking inadequate evaluation. Against a background of concern about standards applied to the evaluation of in vitro diagnostic tests (IVDs), clear statistical thinking was needed on the principles of diagnostic testing in general, and their application in a pandemic. Therefore, in July 2020, the Royal Statistical Society convened a Working Group of six biostatisticians to review the statistical evidence needed to ensure the performance of new tests, especially IVDs for infectious diseases—for regulators, decision-makers, and the public. The Working Group’s review was undertaken when the Covid-19 pandemic shone an unforgiving light on current processes for evaluating and regulating IVDs for infectious diseases. The report’s findings apply more broadly than to the pandemic and IVDs, to diagnostic test evaluations in general. A section of the report focussed on lessons learned during the pandemic and aimed to contribute to the UK Covid-19 Inquiry’s examination of the response to, and impact of, the Covid-19 pandemic to learn lessons for the future. The review made 22 recommendations on what matters for study design, transparency, and regulation.
Differences in ambulatory versus home blood pressure levels by ethnicity: data from the United Kingdom and Japan.
This study tested the hypothesis that differences in ethnicity impact the level of agreement between ambulatory blood pressure (ABP) and home BP (HBP) levels. A retrospective analysis of cross-sectional data from the UK and Japan was performed. Participants underwent office BP, daytime ABP, and HBP measurements. The ABP-HBP difference was compared between ethnic groups by multiple linear regression analysis. Diagnostic disagreement was defined as a disparity between the hypertension diagnoses obtained using ABP and HBP, since both measures share common thresholds of 135/85 mmHg for hypertension. Definite diagnostic disagreement was assigned where such a difference exceeded ±5 mmHg for either systolic BP (SBP) or diastolic BP (DBP). A total of 1 408 participants (age 62.1 ± 11.1 years, 48.6% males, 78.9% known hypertensive, White British 18.9%, South Asian 11.2%, African Caribbean 12.0%, Japanese 58.0%) were eligible. More Japanese participants showed higher ABP than HBP compared to White British: SBP + 3.09 mmHg, 95% confidence interval (CI) + 1.14, +5.04 mmHg; DBP + 5.67 mmHg, 95%CI + 4.51, +6.84 mmHg. More Japanese participants than African Caribbean participants exhibited diagnostic disagreement in SBP (33.2% vs. 20.7%, p = 0.006). Furthermore, Japanese participants had a higher percentage of definite diagnostic disagreement in SBP compared to White British (9.3% vs. 4.5%, p = 0.040) and African Caribbean participants (9.3% vs. 3.0%, p = 0.018). In conclusion, Japanese participants showed greater disparity between ABP and HBP compared to White British participants. Complementary use of ABP and HBP monitoring may be more beneficial for assessing cardiovascular disease risk in Japanese participants compared to other ethnic groups.
Sports-related concussion not associated with long-term cognitive or behavioural deficits: the PROTECT-TBI study.
BACKGROUND: The cognitive effects of sports-related concussion (SRC) have been the subject of vigorous debate but there has been little research into long-term outcomes in non-athlete populations. METHODS: This cohort study of UK community-dwelling adults (aged 50-90 years) was conducted between November 2015 and November 2020, with up to 4 years annual follow-up (n=15 214). Lifetime history of concussions was collected at baseline using the Brain Injury Screening Questionnaire. The first analysis grouped participants by type of concussion (no concussion, only SRC, only non-SRC (nSRC), mixed concussions (both SRC and nSRC)) and the second grouped the participants by number (0, 1, 2 or 3+ SRC or nSRC). Mixed models were used to assess the effect of concussion on outcomes including four cognitive domains and one behavioural measure (Mild Behavioural Impairment-C). RESULTS: Analysis of the included participants (24% male, mean age=64) at baseline found that the SRC group had significantly better working memory (B=0.113, 95% CI 0.038, 0.188) and verbal reasoning (B=0.199, 95% CI 0.092, 0.306) compared with those without concussion. Those who had suffered one SRC had significantly better verbal reasoning (B=0.111, 95% CI 0.031, 0.19) and attention (B=0.115, 95% CI 0.028, 0.203) compared with those with no SRC at baseline. Those with 3+ nSRCs had significantly worse processing speed (B=-0.082, 95% CI -0.144 to -0.019) and attention (B=-0.156, 95% CI -0.248 to -0.063). Those with 3+ nSRCs had a significantly worse trajectory of verbal reasoning with increasing age (B=-0.088, 95% CI -0.149 to -0.026). CONCLUSIONS: Compared with those reporting no previous concussions, those with SRC had no cognitive or behavioural deficits and seemed to perform better in some tasks. As indicated by previous studies, sports participation may confer long-term cognitive benefits.
Changes in intestinal permeability and gut microbiota following diet-induced weight loss in patients with metabolic dysfunction-associated steatohepatitis and liver fibrosis.
Weight loss improves metabolic dysfunction-associated steatohepatitis (MASH). We investigated whether there were associated changes in intestinal permeability, short-chain fatty acids (SCFAs), and gut microbiota, which are implicated in the pathophysiology of MASH. Sixteen adults with MASH, moderate fibrosis, and obesity received a low-energy total diet replacement program for 12 weeks and stepped food re-introduction over the following 12 weeks (ISRCTN12900952). Intestinal permeability, fecal SCFAs, and fecal microbiota were assessed at 0, 12, and 24 weeks. Data were analyzed using mixed-effects linear regression and sparse partial least-squares regression. Fourteen participants completed the trial, lost 15% (95% CI: 11.2-18.6%) of their weight, and 93% had clinically relevant reductions in liver disease severity markers. Serum zonulin concentrations were reduced at both 12 and 24 weeks (152.0 ng/ml, 95% CI: 88.0-217.4, p
Research inefficiencies in external validation studies of the Framingham Wilson coronary heart disease risk rule: A systematic review.
BACKGROUND: External validation studies create evidence about a clinical prediction rule's (CPR's) generalizability by evaluating and updating the CPR in populations different from those used in the derivation, and also by contributing to estimating its overall performance when meta-analysed in a systematic review. While most cardiovascular CPRs do not have any external validation, some CPRs have been externally validated repeatedly. Hence, we examined whether external validation studies of the Framingham Wilson coronary heart disease (CHD) risk rule contributed to generating evidence to their full potential. METHODS: A forward citation search of the Framingham Wilson CHD risk rule's derivation study was conducted to identify studies that evaluated the Framingham Wilson CHD risk rule in different populations. For external validation studies of the Framingham Wilson CHD risk rule, we examined whether authors updated the Framingham Wilson CHD risk rule when it performed poorly. We also assessed the contribution of external validation studies to understanding the Predicted/Observed (P/O) event ratio and c statistic of the Framingham Wilson CHD risk rule. RESULTS: We identified 98 studies that evaluated the Framingham Wilson CHD risk rule; 40 of which were external validation studies. Of these 40 studies, 27 (67.5%) concluded the Framingham Wilson CHD risk rule performed poorly but did not update it. Of 23 external validation studies conducted with data that could be included in meta-analyses, 13 (56.5%) could not fully contribute to the meta-analyses of P/O ratio and/or c statistic because these performance measures were neither reported nor could be calculated from provided data. DISCUSSION: Most external validation studies failed to generate evidence about the Framingham Wilson CHD risk rule's generalizability to their full potential. Researchers might increase the value of external validation studies by presenting all relevant performance measures and by updating the CPR when it performs poorly.
Comparative effects of drug interventions for the acute management of migraine episodes in adults: systematic review and network meta-analysis.
OBJECTIVE: To compare all licensed drug interventions as oral monotherapy for the acute treatment of migraine episodes in adults. DESIGN: Systematic review and network meta-analysis. DATA SOURCES: Cochrane Central Register of Controlled Trials, Medline, Embase, ClinicalTrials.gov, EU Clinical Trials Register, WHO International Clinical Trials Registry Platform, as well as websites of regulatory agencies and pharmaceutical companies without language restrictions until 24 June 2023. METHODS: Screening, data extraction, coding, and risk of bias assessment were performed independently and in duplicate. Random effects network meta-analyses were conducted for the primary analyses. The primary outcomes were the proportion of participants who were pain-free at two hours post-dose and the proportion of participants with sustained pain freedom from two to 24 hours post-dose, both without the use of rescue drugs. Certainty of the evidence was graded using the confidence in network meta-analysis (CINeMA) online tool. Vitruvian plots were used to summarise findings. An international panel of clinicians and people with lived experience of migraine co-designed the study and interpreted the findings. ELIGIBILITY CRITERIA FOR SELECTING STUDIES: Double blind randomised trials of adults (≥18 years) with a diagnosis of migraine according to the International Classification of Headache Disorders. RESULTS: 137 randomised controlled trials comprising 89 445 participants allocated to one of 17 active interventions or placebo were included. All active interventions showed superior efficacy compared with placebo for pain freedom at two hours (odds ratios from 1.73 (95% confidence interval (CI) 1.27 to 2.34) for naratriptan to 5.19 (4.25 to 6.33) for eletriptan), and most of them also for sustained pain freedom to 24 hours (odds ratios from 1.71 (1.07 to 2.74) for celecoxib to 7.58 (2.58 to 22.27) for ibuprofen). In head-to-head comparisons between active interventions, eletriptan was the most effective drug for pain freedom at two hours (odds ratios from 1.46 (1.18 to 1.81) to 3.01 (2.13 to 4.25)), followed by rizatriptan (1.59 (1.18 to 2.17) to 2.44 (1.75 to 3.45)), sumatriptan (1.35 (1.03 to 1.75) to 2.04 (1.49 to 2.86)), and zolmitriptan (1.47 (1.04 to 2.08) to 1.96 (1.39 to 2.86)). For sustained pain freedom, the most efficacious interventions were eletriptan and ibuprofen (odds ratios from 1.41 (1.02 to 1.93) to 4.82 (1.31 to 17.67)). Confidence in accordance with CINeMA ranged from high to very low. Sensitivity analyses on Food and Drug Administration licensed doses only, high versus low doses, risk of bias, and moderate to severe headache at baseline confirmed the main findings for both primary and secondary outcomes. CONCLUSIONS: Overall, eletriptan, rizatriptan, sumatriptan, and zolmitriptan had the best profiles and they were more efficacious than the recently marketed drugs lasmiditan, rimegepant, and ubrogepant. Although cost effectiveness analyses are warranted and careful consideration should be given to patients with a high risk cardiovascular profile, the most effective triptans should be considered as preferred acute treatment for migraine and included in the WHO List of Essential Medicines to promote global accessibility and uniform standards of care. SYSTEMATIC REVIEW REGISTRATION: Open Science Framework https://osf.io/kq3ys/.
Innovations in blood pressure measurement and reporting technology: International Society of Hypertension position paper endorsed by the World Hypertension League, European Society of Hypertension, Asian Pacific Society of Hypertension, and Latin American Society of Hypertension.
Blood pressure (BP) is a key contributor to the lifetime risk of preclinical organ damage and cardiovascular disease. Traditional clinic-based BP readings are typically measured infrequently and under standardized/resting conditions and therefore do not capture BP values during normal everyday activity. Therefore, current hypertension guidelines emphasize the importance of incorporating out-of-office BP measurement into strategies for hypertension diagnosis and management. However, conventional home and ambulatory BP monitoring devices use the upper-arm cuff oscillometric method and only provide intermittent BP readings under static conditions or in a limited number of situations. New innovations include technologies for BP estimation based on processing of sensor signals supported by artificial intelligence tools, technologies for remote monitoring, reporting and storage of BP data, and technologies for BP data interpretation and patient interaction designed to improve hypertension management ("digital therapeutics"). The number and volume of data relating to new devices/technologies is increasing rapidly and will continue to grow. This International Society of Hypertension position paper describes the new devices/technologies, presents evidence relating to new BP measurement techniques and related indices, highlights standard for the validation of new devices/technologies, discusses the reliability and utility of novel BP monitoring devices, the association of these metrics with clinical outcomes, and the use of digital therapeutics. It also highlights the challenges and evidence gaps that need to be overcome before these new technologies can be considered as a user-friendly and accurate source of novel BP data to inform clinical hypertension management strategies.
Experiences of integrating social prescribing link workers into primary care in England: Bolting on, fitting in or belonging.
BACKGROUND: Following the 2019 NHS Long-Term Plan, link workers (LWs) have been employed across primary care in England to deliver social prescribing (SP). AIM: To understand and explain how the LW role is being implemented in primary care in England. DESIGN AND SETTING: Realist evaluation undertaken in England. METHOD: Focused ethnographies around seven LWs from different parts of England. As part of this, we interviewed 61 patients and 93 professionals from healthcare and the voluntary, community and social enterprise (VCSE) sector. We reinterviewed 41 patients, seven LWs and a LW manager 9-12 months after their first interview. RESULTS: We developed four concepts around how LWs are integrated (or not) within primary care: Centralising or diffusing power; Forging an identity in general practice; Demonstrating effect; Building a facilitative infrastructure. These concepts informed the development of a programme theory around a continuum of integration of LWs into primary care - from being 'bolted on' to existing provision, without much consideration, to 'fitting in', shaping what is delivered to be accommodating, through to 'belonging', whereby they are accepted as a legitimate source of support, making a valued contribution to patients' broader well-being. CONCLUSION: SP was introduced into primary care to promote greater attention to the full range of factors affecting patients' health and well-being, beyond biomedicine. For that to happen, our analysis highlights the need for a whole system approach to defining, delivering and maintaining this new part of practice.
Physical, cognitive, and social triggers of symptom fluctuations in people living with long COVID: an intensive longitudinal cohort study
Background: Symptom fluctuations within and between individuals with long COVID are widely reported, but the extent to which severity varies following different types of activity and levels of exertion, and the timing of symptoms and recovery, have not previously been quantified. We aimed to characterise timing, severity, and nature of symptom fluctuations in response to effortful physical, social and cognitive activities, using Ecological Momentary Assessments. Methods: We recorded activity, effort, and severity of 8 core symptoms every 3 h for up to 24 days, in cohorts from both clinic and community settings. Symptom severities were jointly modelled using autoregressive and moving average processes. Findings: Consent was received from 376 participants providing ≥1 week's measurements (273 clinic-based, 103 community-based). Severity of all symptoms was elevated 30 min after all categories of activity. Increased effort was associated with increased symptom severity. Fatigue severity scores increased by 1.8/10 (95% CI: 1.6–1.9) following the highest physical exertions and by 1.5 (1.4–1.7) following cognitive efforts. There was evidence of only mild delayed fatigue 3 h (0.3, 0.2–0.5) or one day later (0.2, 0.0– 0.5). Fatigue severity increased as the day progressed (1.4, 1.0–1.7), and cognitive dysfunction was 0.2 lower at weekends (0.1–0.3). Interpretation: Cognitive, social, self-care and physical activities all triggered increased severity across every symptom, consistent with associated common pathways as potential therapeutic targets. Clear patterns of symptom fluctuations emerged that support more targeted self-management. Funding: National Institute for Health and Care Research.