Alan Watkins’s research while affiliated with Swansea University and other places

Background A new lymphoedema-specific Patient Reported Outcome Measure (LYMPROM©) was developed to help patients easily report the impact of their lymphoedema and enable lymphoedema therapists to understand what matters most to patients. Aims This paper describes the validation of LYMPROM© for adults with lymphoedema. Methods A multi-phased iterative review was undertaken to investigate the reliability and validity of LYMPROM©. Face and content validity were reviewed by surveying patient representatives and healthcare professionals, along with the validity of LYMPROM© Cymraeg, a Welsh translation. Following COSMIN guidelines, validation study phases used anonymised routinely collected data to examine internal consistency, structural validity, construct validity (compared with the EQ5D-5L), measurement error, test-retest reliability and responsiveness. Results LYMPROM© demonstrated validity (content and construct) and reliability (test-retest, internal consistency). All items were regarded as relevant, comprehensive and clear, with item content validity index (CVI) between 0.83 to 1.00, and average overall assessment of 0.94. Robust development of LYMPROM© Cymraeg ensured appropriate translation into Welsh. LYMPROM© item scores, with means and medians generally in the lower half of the scale, were positively correlated, as were three (Physical health, Social health and Emotional health) domain scores (domain correlations: 0.595 to 0.812). LYMPROM© total and domain scores showed moderate negative correlations (-0.577 to -0.435) with EQ5D-5L measures. LYMPROM© total and domain scores showed good test-retest (within two weeks) properties, with little or no change in mean or median scores, and strong positive correlations between test and retest scores (Total: 0.919; Physical health domain: 0.922; Social health domain: 0.889; Emotional health domain: 0.820). LYMPROM© showed good responsiveness, with strong, positive correlations between total and domain initial and repeat (between four weeks and seven months later) scores, with a slight reduction in scores (-3.8 to -2.0 units) and some indication of relationships between reduction and time interval (Total: p = 0.025; Physical health domain: 0.034; Social health domain: 0.181; Emotional health domain: 0.009). Conclusion Evidence shows that LYMPROM© offers a reliable and valid tool for use in clinical practice. Scores on three domains allow a more granular assessment of the patient’s view of their condition; these scores and the total LYMPROM© score exhibit moderate correlations with more generic EQ5D-5L measures. Further research will explore relationships between patient-level characteristics and LYMPROM© responses, and extend initial work on its cross-cultural validity.

This observational cohort study explored lymphoedema development following a cancer diagnosis and whether demographic factors impacted the time to lymphoedema development. We identified cases through the Secure Anonymised Information Linkage (SAIL) Databank. We used cancer diagnostic codes to identify a cohort of six broad cancer ‘types’. We independently used lymphoedema diagnostic codes to identify a cohort who developed lymphoedema. We linked these two cohorts to develop a single cohort of cases and describe the number of cases who went on to develop lymphoedema after a cancer diagnosis, and the time to lymphoedema diagnosis. We used Cox regression models to calculate hazard ratios and produced survival curves to explore whether pre‐defined factors (gender, age, deprivation, cancer type) had any impact on time to lymphoedema development. We identified 7538 cases of lymphoedema development after a cancer diagnosis, relating to 7279 people. There was considerable variation in the time to diagnosis, with a mean and standard deviation of 483.3 (701.8) days. Cancer type was the single most important factor in explaining time to lymphoedema diagnosis. Time to lymphoedema was shortest in breast cancer. A large number of breast cancer cases have undergone surgery, and this may account for the earlier development of lymphoedema. Consideration should be made of risk factors for lymphoedema development in order to allow for more targeted treatment plans that could improve health‐related quality of life for patients.

Aim Incisional Hernia (IH) is a common complication of colorectal surgery, affecting up to 30% of patients. There is an association between surgeon experience and outcomes in hernia surgery, yet little regarding impact of the surgeon performing abdominal wall closure (AWC) on IH rate. We aimed to assess the rates of IH at 1 year following AWC between junior and senior surgeons in patients undergoing colorectal surgery. Method This is an exploratory analysis of patients who participated in the Hughes Abdominal Repair Trial (HART), a prospective, multicentre randomised control trial comparing AWC methods. Grade of surgeon performing AWC was categorised into ‘Trainee’ and ‘Consultant’ and compared to IH rate at 1 year. Results 663 patients were included. The rate of IH in patients closed by trainees was 20%, compared to 12% in those closed by consultants (P = <0.001). Patients closed by trainees were more likely to have had rectal surgery. Patients closed by consultants were more likely to be younger and have a longer duration of surgery. On multivariate analysis, age, male sex and closure by a trainee were identified as risk factors for developing IH. Conclusion Patients who undergo AWC by a trainee have an increased risk of developing IH when compared to those closed by a consultant. Further work is needed to determine the impact of supervised and un-supervised trainees on IH rates. AWC should be regarded as a training opportunity in its own right, and closure time should be seen as training time, not coffee time.

Background This study evaluated the impact of 2015/2016 prescribing guidance on antidepressant prescribing choices in children. Methods A retrospective e-cohort study of whole population routine electronic healthcare records was conducted. Poisson regression was undertaken to explore trends over time for depression, antidepressant prescribing, indications and secondary care contacts. Time trend analysis was conducted to assess the impact of guidance. Results A total of 643 322 primary care patients in Wales UK, aged 6–17 years from 2010–2019 contributed 3 215 584 person-years of follow-up. Adjusted incidence of depression more than doubled (IRR for 2019 = 2.8 [2.5–3.2]) with similar trends seen for antidepressants. Fluoxetine was the most frequently prescribed first-line antidepressant. Citalopram comprised less than 5% of first prescriptions in younger children but 22.9% (95% CI 22.0–23.8; 95% CI 2533) in 16–17-year-olds. Approximately half of new antidepressant prescribing was associated with depression. Segmented regression analysis showed that prescriptions of ‘all’ antidepressants, Fluoxetine and Sertraline were increasing before the guidance. This upward trend flattened for both ‘all’ antidepressants and Fluoxetine and steepened for Sertraline. Citalopram prescribing was decreasing significantly pre guidance being issued with no significant change afterward. Conclusions Targeted intervention is needed to address rising rates of depression in children. Practitioners are partially adhering to local and national guidance. The decision-making process behind prescribing choices is likely to be multi-factorial. Activities to support implementation of guidance should be adopted in relation to safety in prescribing of antidepressants in children including timely availability of talking therapies and specialist mental health services.

Background Busy periods, such as ‘winter pressures’ months, can create challenges for Emergency Departments (EDs) managing patient flow. This may increase risks throughout the healthcare system. ED patients may receive suboptimal care, some patients may remain in ambulances, sometimes for hours, whilst queued ambulances cannot attend other patients.In some EDs, ambulance queueing is relatively rare; in others, it is more common. As part of the STALLED study, we investigated any association between mean ambulance handover time and the number of monthly handovers. Methods We analysed publicly available ambulance collection data for English NHS Trusts between October 2023 and March 2024 from NHS England. We included all Type 1 Acute Trusts, excluding children’s hospitals, those with fewer than 100 handovers per month, and clear outliers. Results 105 Trusts were included (10 to 18 per English region). The number of handovers recorded per month varied between 716 and 8,404 with a mean of 3,090. Monthly mean handover time varied between 8 minutes, 45 seconds and 129 minutes, 6 seconds. Figure 1 shows a weak relation between mean handover time and mean monthly handovers. • Download figure • Open in new tab • Download powerpoint Abstract PP35 Figure 1 Conclusion Mean ambulance handover time is not obviously correlated with mean monthly number of handovers. Therefore, we propose the existence of deeper-rooted obstacles/challenges which warrant further exploration. It also remains to assess temporal patterns in more detail.While queueing is a problem everywhere to some extent, there is variation in how EDs manage it. Understanding these variations may lead to improvements in patient safety, health outcomes, experience, and costs.

Background Multi-agency case management across the UK addresses the needs of those who call 999 ambulance services frequently (≥5 times in a month, or ≥12 times in a three-month period). However, there is little evidence about what works and how. STRETCHED (Strategies to Manage Emergency Ambulance Telephone Callers with Sustained High needs – an Evaluation using Linked Data) evaluated case management using mixed methods. One objective was to develop and refine a logic model to describe cross-sectoral case management and its expected impacts, as a basis for evaluation. Methods Stage 1: We conducted a stakeholder event with 37 people from Wales, England and Northern Ireland including public partners (with lived experience) and relevant staff to develop an initial logic model.Stage 2: Following the main data collection and analysis phases of the study, we reviewed and refined the logic model in the light of findings. Results The logic model summarises in graphic form: • Inputs resources and activities • Mechanisms of change at organisational and individual level • Impacts on service users, ambulance services, individual staff, and wider service networks. Study findings showed that: case management initiatives varied considerably between sites; patients had a high level of need with high mortality and rates of hospital admission; the conditions of many patients are not amenable to individual behaviour change. We then revised the logic model to show: • the range of contextual/mediating factors shaping local initiatives • variation in case management activities between sites • a reduced emphasis on individual behaviour change. Conclusion In this challenging and complex area of health care, a logic model can be an essential tool to support evaluation, and refined on the basis of study insights.

Introduction Hospital-based violence intervention programmes (HVIPs), based in Emergency Departments (EDs) have been proposed as a public health response to violence. These programmes address the underlying reasons why patients are exposed to violence. In addressing any underlying modifiable risks and vulnerabilities HVIPs can reduce patients’ exposure to violence and therefore subsequent unplanned attendance into ED. Methods and Analysis ED patients are eligible for inclusion in the evaluation if they are normally resident in Wales, United Kingdom (UK), aged 11 years and older. A controlled longitudinal natural experiment will be undertaken. The primary outcome is derived from the Emergency Department Dataset, routinely collected for all EDs in Wales, and is subsequent unplanned ED attendance. Case patients will be matched to control patients attending EDs without an HVIP. Analysis will derive the hazard rate for subsequent unplanned ED attendances using recurrent event analysis. The total monthly count of patients identified as attending because of violence in intervention EDs will be compared to the total count of Welsh control EDs in an interrupted time series analysis to determine whether HVIPS increase violence ascertainment. To determine whether referral, versus no referral, to the HVIP represents value for money, we will undertake a cost-effectiveness analysis from the perspective of the National Health Service. Ethics and Dissemination The approval to access and analyse data housed in the Secure Anonymised Information Linkage (SAIL) databank, an ISO 27001 certified and UK Statistics Authority accredited secure data environment, was granted by the SAIL independent Information Governance Review Panel (Ref: 1421). Findings will be presented at local, national, and international conferences and disseminated by peer-review publication. ISRCTN Registration : 41868 (12 August 2022)

Background Early assessment of patients with suspected transient ischaemic attack (TIA) is crucial to provision of effective care, including initiation of preventive therapies and identification of stroke mimics. Many patients with TIA present to emergency medical services (EMS) but may not require hospitalisation. Paramedics could identify and refer patients with low-risk TIA, without conveyance to the ED. Safety and effectiveness of this model is unknown. Aim To assess the feasibility of undertaking a fully powered randomised controlled trial (RCT) to evaluate clinical and cost-effectiveness of paramedic referral of patients who call EMS with low-risk TIA to TIA clinic, avoiding transfer to ED. Methods The Transient Ischaemic attack Emergency Referral (TIER) intervention was developed through a survey of UK ambulance services, a scoping review of evidence of prehospital care of TIA and convening a specialist clinical panel to agree its final form. Paramedics in South Wales, UK, were randomly allocated to trial intervention (TIA clinic referral) or control (usual care) arms, with patients’ allocation determined by that of attending paramedics. Predetermined progression criteria considered: proportion of patients referred to TIA clinic, data retrieval, patient satisfaction and potential cost-effectiveness. Results From December 2016 to September 2017, eighty-nine paramedics recruited 53 patients (36 intervention; 17 control); 48 patients (31 intervention; 17 control) consented to follow-up via routine data. Three intervention patients, of seven deemed eligible, were referred to TIA clinic by paramedics. Contraindications recorded for the other intervention arm patients were: Face/Arms/Speech/Time positive (n=13); ABCD2 score >3 (n=5); already anticoagulated (n=2); crescendo TIA (n=1); other (n=8). Routinely collected electronic health records, used to report further healthcare contacts, were obtained for all consenting patients. Patient-reported satisfaction with care was higher in the intervention arm (mean 4.8/5) than the control arm (mean 4.2/5). Health economic analysis suggests an intervention arm quality-adjusted life-year loss of 0.0094 (95% CI −0.0371, 0.0183), p=0.475. Conclusion The TIER feasibility study did not meet its progression criteria, largely due to low patient identification and referral rates. A fully powered RCT in this setting is not recommended. Trial registration number ISRCTN85516498 .