Free-text Data Research Articles

User profile of people contacting a stroke helpline (StrokeLine) in Australia: a retrospective cohort study

Background StrokeLine is a specialised telephone helpline led by health professionals in Australia. Aims (i) To describe the profile of StrokeLine callers; (ii) to understand the reasons people engage with the service and (iii) how StrokeLine responded to the caller's needs. Methods Routine call data were obtained from the StrokeLine between November 2019 and November 2020. Data were extracted and descriptive analyses performed. De-identified free-text data were obtained separately for November 2019 and June 2020 and analysed using qualitative content analysis. Results Of the 1429 calls most were from carers, family and friends (38%) or the stroke survivor themselves (34%). Most calls were made by women (64%) and the average age of the stroke survivor was ≥65 years (33%) with the time since the stroke occurred <1 year. The main reason for calling was to manage stroke-related impairments (40%). Providing information, support and advice was the most common action provided by StrokeLine staff (25%). Content analysis of 225 calls revealed most stroke survivors called for emotional support, while carers sought more practical guidance. StrokeLine provided information for referral to relevant services and guidance on what to do next. Conclusions Most calls were received from family and carers, as well as stroke survivors. They contacted StrokeLine for information and advice, practical solutions, emotional support, and referral advice to other services.

Impacts of an Electronic Health Record Transition on Veterans Health Administration Health Professions Trainee Experience.

Adoption of electronic health care records (EHRs) has proliferated since 2000. While EHR transitions are widely understood to be disruptive, little attention has been paid to their effect on health professions trainees' (HPTs) ability to learn and conduct work. Veterans Health Administration's (VA) massive transition from its homegrown EHR (CPRS/Vista) to the commercial Oracle Cerner presents an unparalleled-in-scope opportunity to gain insight on trainee work functions and their ability to obtain requisite experience during transitions. To identify how an organizational EHR transition affected HPT work and learning at the third VA go-live site. A formative mixed-method evaluation of HPT experiences with VHA's EHR transition including interviews with HPTs and supervisors at Chalmers P. Wylie VA Outpatient Clinic in Columbus, OH, before (~60 min), during (15-30 min), and after (~60 min) go-live (December 2021-July 2022). We also conducted pre- (March 2022-April 2022) and post-go live (May 2022-June 2022) HPT and employee surveys. We conducted 24 interviews with HPTs (n=4), site leaders (n=2), and academic affiliates (n=2) using snowball sampling. We recruited HPTs in pre- (n=13) and post-go-live (n=10) surveys and employees in pre- (n=408) and post-go-live (n=458) surveys. We conducted interviews using a semi-structured guide and grounded prompts. We coded interviews and survey free text data using a priori and emergent codes, subsequently conducting thematic analysis. We conducted descriptive statistical analysis of survey responses and merged interview and survey data streams. Our preliminary findings indicate that the EHR transition comprehensively affected HPT experiences, disrupting processes from onboarding and training to clinical care contributions and training-to-career retention. Understanding HPTs' challenges during EHR transitions is critical to effective training. Mitigating the identified barriers to HPT training and providing patient care may lessen their dissatisfaction and ensure quality patient care during EHR transitions.

Benefits of Mentoring in Oncology Education for Mentors and Mentees: Pre-Post Interventional Study of the British Oncology Network for Undergraduate Societies' National Oncology Mentorship Scheme.

Formal education of oncology is lacking in many undergraduate medical curricula. Mentoring schemes can expose participants to specific areas of medicine and may address the shortfalls in oncology education. Few mentoring schemes have been designed within the United Kingdom, especially within oncology. There is a need to understand reasons for mentor and mentee participation in such schemes and to identify ways to minimize barriers to engagement. This study identifies motivations for participation in an oncology mentoring scheme and its benefits and limitations to both the mentee and the mentor. The British Oncology Network for Undergraduate Societies launched a National Oncology Mentorship Scheme (NOMS) on September 1, 2021. Mentees (medical student or foundation doctor) were paired with mentors (specialty registrar or consultant), for 6 months of mentoring. In total, 86 mentors and 112 mentees were recruited to the scheme. The mentees and mentors were asked to meet at least 3 times during this period and suggestions were provided on the content of mentoring. Mentees and mentors were invited to complete a prescheme questionnaire, exploring motivations for involvement in the scheme, current experiences within oncology, and knowledge and interests in the field. At the end of the scheme, mentors and mentees were asked to complete a postscheme questionnaire exploring experiences and benefits or limitations of participation. Paired analysis was performed using the Wilcoxon signed-rank test. For free text data, content analysis was applied to summarize the main themes in the data. Of the 66 (59%) mentees who completed the prescheme questionnaire, 41 (62%) were clinical, 21 (32%) preclinical medical students, and the remainder were junior doctors. For mentees, networking was the primary reason for joining the scheme (n=25, 38%). Mentees ranked experience of oncology at medical school at 3 on 10 (IQR 2-5). In this, 46 (53%) mentors completed the prescheme questionnaire, 35 (76%) were registrar level, and the remainder were consultant level (n=11). The most common reason for mentor participation was to increase awareness and interest in the field (n=29, 63%). Of those who completed the prescheme questionnaire, 23 (35%) mentees and 25 (54%) mentors completed the postscheme questionnaire. Knowledge in all areas of oncology assessed significantly increased during the scheme (P<.001). Most mentees (n=21, 91%) and mentors (n=18, 72%) felt they had benefited from the scheme. Mentees cited gaining insights into oncology as most beneficial; and mentors, opportunities to develop professionally. Whilst mentees did not report any barriers to participating in the scheme, mentors stated lack of time as the greatest barrier to mentoring. British Oncology Network for Undergraduate Societies' NOMS is expanding and is beneficial for mentees through increasing knowledge, providing exposure, and career advice in oncology. Mentors benefit from improving their mentoring skills and personal satisfaction.

Development of machine learning models for the detection of surgical site infections following total hip and knee arthroplasty: a multicenter cohort study

BackgroundPopulation based surveillance of surgical site infections (SSIs) requires precise case-finding strategies. We sought to develop and validate machine learning models to automate the process of complex (deep incisional/organ space) SSIs case detection.MethodsThis retrospective cohort study included adult patients (age ≥ 18 years) admitted to Calgary, Canada acute care hospitals who underwent primary total elective hip (THA) or knee (TKA) arthroplasty between Jan 1st, 2013 and Aug 31st, 2020. True SSI conditions were judged by the Alberta Health Services Infection Prevention and Control (IPC) program staff. Using the IPC cases as labels, we developed and validated nine XGBoost models to identify deep incisional SSIs, organ space SSIs and complex SSIs using administrative data, electronic medical records (EMR) free text data, and both. The performance of machine learning models was assessed by sensitivity, specificity, positive predictive value, negative predictive value, F1 score, the area under the receiver operating characteristic curve (ROC AUC) and the area under the precision–recall curve (PR AUC). In addition, a bootstrap 95% confidence interval (95% CI) was calculated.ResultsThere were 22,059 unique patients with 27,360 hospital admissions resulting in 88,351 days of hospital stay. This included 16,561 (60.5%) TKA and 10,799 (39.5%) THA procedures. There were 235 ascertained SSIs. Of them, 77 (32.8%) were superficial incisional SSIs, 57 (24.3%) were deep incisional SSIs, and 101 (42.9%) were organ space SSIs. The incidence rates were 0.37 for superficial incisional SSIs, 0.21 for deep incisional SSIs, 0.37 for organ space and 0.58 for complex SSIs per 100 surgical procedures, respectively. The optimal XGBoost models using administrative data and text data combined achieved a ROC AUC of 0.906 (95% CI 0.835–0.978), PR AUC of 0.637 (95% CI 0.528–0.746), and F1 score of 0.79 (0.67–0.90).ConclusionsOur findings suggest machine learning models derived from administrative data and EMR text data achieved high performance and can be used to automate the detection of complex SSIs.

“Pouring their heart out in Sainsbury’s”: qualitative study of young people’s, parents’ and mental health practitioners’ experiences of adapting to remote online mental health appointments during COVID-19

Background During the COVID-19 pandemic, technologies such as videoconferencing were used to deliver mental health appointments remotely online. For many people, this was a change from previous methods of mental healthcare receipt and delivery. We aimed to explore in-depth how practitioners, young people and parents in the UK experienced this transition.MethodsWe used qualitative methods to collect data, triangulating between free-text online survey data (n = 38), focus groups (n = 5) (3 young adult groups (total n = 11); 2 practitioner groups (total n = 7)), and semi-structured interviews (practitioners n = 8; parents n = 4). Data were analysed using reflexive thematic analysis.ResultsParticipants held mixed views about remote appointments, which were encompassed within the five themes of: home as clinic; disrupted therapeutic relationships; difficulties with engagement; uncontained risk; and scope of care provision. While appointments at home could be regarded as more comfortable, naturalistic and accessible, it was also recognised that remoteness compromised practitioner control with consequences for their ability to monitor patient engagement, manage risk and ensure confidentiality when others were present in the home. This could create an additional burden for parents as they tried to facilitate appointments but felt unsupported in this role. Relatedly, remoteness was seen to hinder interpersonal communication, formation of trust, communication of empathy and opportunities to observe body language, all of which were deemed important to building and maintaining effective therapeutic relationships. Despite this, others thought the anonymity of a remote exchange may allow earlier disclosure. There was disagreement as to whether remote provision narrowed or expanded the scope of practice.ConclusionsWhile some had positive views of remote mental health appointments, others found them challenging. Findings highlight key areas requiring attention and mitigation in future offerings of remote provision, namely: risk management, parental burden, and problematic engagement.

An interpretable deep learning framework for predicting liver metastases in postoperative colorectal cancer patients using natural language processing and clinical data integration.

The significance of liver metastasis (LM) in increasing the risk of death for postoperative colorectal cancer (CRC) patients necessitates innovative approaches to predict LM. Our study presents a novel and significant contribution by developing an interpretable fusion model that effectively integrates both free-text medical record data and structured laboratory data to predict LM in postoperative CRC patients. We used a robust dataset of 1463 patients and leveraged state-of-the-art natural language processing (NLP) and machine learning techniques to construct a two-layer fusion framework that demonstrates superior predictive performance compared to single modal models. Our innovative two-tier algorithm fuses the results from different data modalities, achieving balanced prediction results on test data and significantly enhancing the predictive ability of the model. To increase interpretability, we employed Shapley additive explanations to elucidate the contributions of free-text clinical data and structured clinical data to the final model. Furthermore, we translated our findings into practical clinical applications by creating a novel NLP score-based nomogram using the top 13 valid predictors identified in our study. The proposed fusion models demonstrated superior predictive performance with an accuracy of 80.8%, precision of 80.3%, recall of 80.5%, and an F1 score of 80.8% in predicting LMs. This fusion model represents a notable advancement in predicting LMs for postoperative CRC patients, offering the potential to enhance patient outcomes and support clinical decision-making.

Young people’s opinions of cancer care in England: the BRIGHTLIGHT cohort

ObjectivesThe BRIGHTLIGHT cohort study was the national evaluation of cancer services for teenager and young adults (TYA). This was analysis of free-text survey data to better understand their experiences of...

Using Clinician-Patient WeChat Group Communication Data to Identify Symptom Burdens in Patients With Uterine Fibroids Under Focused Ultrasound Ablation Surgery Treatment: Qualitative Study.

Unlike research project-based health data collection (questionnaires and interviews), social media platforms allow patients to freely discuss their health status and obtain peer support. Previous literature has pointed out that both public and private social platforms can serve as data sources for analysis. This study aimed to use natural language processing (NLP) techniques to identify concerns regarding the postoperative quality of life and symptom burdens in patients with uterine fibroids after focused ultrasound ablation surgery. Screenshots taken from clinician-patient WeChat groups were converted into free texts using image text recognition technology and used as the research object of this study. From 408 patients diagnosed with uterine fibroids in Chongqing Haifu Hospital between 2010 and 2020, we searched for symptom burdens in over 900,000 words of WeChat group chats. We first built a corpus of symptoms by manually coding 30% of the WeChat texts and then used regular expressions in Python to crawl symptom information from the remaining texts based on this corpus. We compared the results with a manual review (gold standard) of the same records. Finally, we analyzed the relationship between the population baseline data and conceptual symptoms; quantitative and qualitative results were examined. A total of 408 patients with uterine fibroids were included in the study; 190,000 words of free text were obtained after data cleaning. The mean age of the patients was 39.94 (SD 6.81) years, and their mean BMI was 22.18 (SD 2.78) kg/m2. The median reporting times of the 7 major symptoms were 21, 26, 57, 2, 18, 30, and 49 days. Logistic regression models identified preoperative menstrual duration (odds ratio [OR] 1.14, 95% CI 5.86-6.37; P=.009), age of menophania (OR -1.02 , 95% CI 11.96-13.47; P=.03), and the number (OR 2.34, 95% CI 1.45-1.83; P=.04) and size of fibroids (OR 0.12, 95% CI 2.43-3.51; P=.04) as significant risk factors for postoperative symptoms. Unstructured free texts from social media platforms extracted by NLP technology can be used for analysis. By extracting the conceptual information about patients' health-related quality of life, we can adopt personalized treatment for patients at different stages of recovery to improve their quality of life. Python-based text mining of free-text data can accurately extract symptom burden and save considerable time compared to manual review, maximizing the utility of the extant information in population-based electronic health records for comparative effectiveness research.

614 “Women in Surgery – Breaking Stereotypes”: Value of Organising WINS Sessions Early in Medical Teaching to Tackle Stereotypes

Abstract Aim Despite 59% medical students being female, only 12% consultant surgeons in the UK are female. This study explores the value offered by teaching sessions to promote Women in Surgery (WINS) and help tackle stereotypes amongst medical students. Method A virtual session aimed at medical students delivered by a female surgeon who shared her journey of being and becoming a surgeon was offered. A post-session survey collected data from the attendees on their perceptions of the challenges women face in pursuing a career in surgery. Free-text data from the survey were analysed thematically. Results Three key themes emerged based on the views attendees shared on the challenges women face in surgery– ‘women cannot make good surgeons’; ‘surgery is harder for females to access’ and ‘women are more family-orientated with less passion for the job’. Following the session, the participants were asked to share their key learning points which they identified as ‘women can make excellent surgeons’; ‘support is available’ and ‘work-life balance is possible’. Overall, all participants stated that the session motivated them in taking up a career in surgery. Conclusions Our results indicate that direct early medical student-to-surgeon engagement is significant in addressing misconceptions which might be helpful in dealing with some stereotypes surrounding WINS. Well-informed students are more likely to pursue surgery as a career, addressing current gender disparities. Feedback indicated that attendees valued personal insights and experiences of the invited speaker and this information is of value to inform future interventions which help address issues of representation of females in surgery.

1009 Improving Accuracy and Extractability of Electronic Operative Documentation Data in a Hand Trauma Unit

Abstract Aim Accurate and thorough operative documentation is essential for continuity of patient care, audit purposes, and the medicolegal protection of patients and doctors. In a hand trauma department in central London, challenges to this include a fast turnover of junior doctors, busy trauma lists, and complex, multidisciplinary cases. ‘Pathpoint’ is an online system with functionality for operative note templates and pre-filled data. However, it is unclear how much this is utilised in the department. This project aimed to evaluate operative documentation for hand trauma and the utilisation of templates for data input. Method All hand trauma cases performed in a one-month period were requested from Pathpoint. Operation notes were initially screened for directly extractable data in template-form then manually checked for free-text data entry using prior-agreed criteria. Results The initial search yielded 51 procedures. Nerve repairs were best documented (18% data missing), terminalisation procedures were worst documented (38% data missing). Overall, only 18.3% of all data were directly extractable; 0% data were extractable for dressings, post-operative antibiotics, follow-up and movement restrictions. Best documented topics were hand trauma findings, skin closure, and follow-up. &amp;gt;90% of data in each of these were freetext. Worst documented topics were CEPOD category, ASA grade, intra-operative antibiotics, and tourniquet details. Conclusions Despite overall good documentation, multiple important omissions are noted. Furthermore, the majority of data is not recorded in an extractable format. Templates have been distributed within the department. Re-auditing in subsequent months will assess whether accuracy of documentation and ease of data extraction have improved.

Web-Based Application Based on Human-in-the-Loop Deep Learning for Deidentifying Free-Text Data in Electronic Medical Records: Development and Usability Study.

The narrative free-text data in electronic medical records (EMRs) contain valuable clinical information for analysis and research to inform better patient care. However, the release of free text for secondary use is hindered by concerns surrounding personally identifiable information (PII), as protecting individuals' privacy is paramount. Therefore, it is necessary to deidentify free text to remove PII. Manual deidentification is a time-consuming and labor-intensive process. Numerous automated deidentification approaches and systems have been attempted to overcome this challenge over the past decade. We sought to develop an accurate, web-based system deidentifying free text (DEFT), which can be readily and easily adopted in real-world settings for deidentification of free text in EMRs. The system has several key features including a simple and task-focused web user interface, customized PII types, use of a state-of-the-art deep learning model for tagging PII from free text, preannotation by an interactive learning loop, rapid manual annotation with autosave, support for project management and team collaboration, user access control, and central data storage. DEFT comprises frontend and backend modules and communicates with central data storage through a filesystem path access. The frontend web user interface provides end users with a user-friendly workspace for managing and annotating free text. The backend module processes the requests from the frontend and performs relevant persistence operations. DEFT manages the deidentification workflow as a project, which can contain one or more data sets. Customized PII types and user access control can also be configured. The deep learning model is based on a Bidirectional Long Short-Term Memory-Conditional Random Field (BiLSTM-CRF) with RoBERTa as the word embedding layer. The interactive learning loop is further integrated into DEFT to speed up the deidentification process and increase its performance over time. DEFT has many advantages over existing deidentification systems in terms of its support for project management, user access control, data management, and an interactive learning process. Experimental results from DEFT on the 2014 i2b2 data set obtained the highest performance compared to 5 benchmark models in terms of microaverage strict entity-level recall and F1-scores of 0.9563 and 0.9627, respectively. In a real-world use case of deidentifying clinical notes, extracted from 1 referral hospital in Sydney, New South Wales, Australia, DEFT achieved a high microaverage strict entity-level F1-score of 0.9507 on a corpus of 600 annotated clinical notes. Moreover, the manual annotation process with preannotation demonstrated a 43% increase in work efficiency compared to the process without preannotation. DEFT is designed for health domain researchers and data custodians to easily deidentify free text in EMRs. DEFT supports an interactive learning loop and end users with minimal technical knowledge can perform the deidentification work with only a shallow learning curve.

Machine Learning-Enabled Clinical Information Systems Using Fast Healthcare Interoperability Resources Data Standards: Scoping Review.

Machine learning-enabled clinical information systems (ML-CISs) have the potential to drive health care delivery and research. The Fast Healthcare Interoperability Resources (FHIR) data standard has been increasingly applied in developing these systems. However, methods for applying FHIR to ML-CISs are variable. This study evaluates and compares the functionalities, strengths, and weaknesses of existing systems and proposes guidelines for optimizing future work with ML-CISs. Embase, PubMed, and Web of Science were searched for articles describing machine learning systems that were used for clinical data analytics or decision support in compliance with FHIR standards. Information regarding each system's functionality, data sources, formats, security, performance, resource requirements, scalability, strengths, and limitations was compared across systems. A total of 39 articles describing FHIR-based ML-CISs were divided into the following three categories according to their primary focus: clinical decision support systems (n=18), data management and analytic platforms (n=10), or auxiliary modules and application programming interfaces (n=11). Model strengths included novel use of cloud systems, Bayesian networks, visualization strategies, and techniques for translating unstructured or free-text data to FHIR frameworks. Many intelligent systems lacked electronic health record interoperability and externally validated evidence of clinical efficacy. Shortcomings in current ML-CISs can be addressed by incorporating modular and interoperable data management, analytic platforms, secure interinstitutional data exchange, and application programming interfaces with adequate scalability to support both real-time and prospective clinical applications that use electronic health record platforms with diverse implementations.

Salient Measures of Hospitalist Workload

The ideal hospitalist workload and optimal way to measure it are not well understood. To obtain expert consensus on the salient measures of hospitalist workload. This qualitative study used a 3-round Delphi technique between April 5 and July 13, 2022, involving national experts within and external to the field. Experts included hospitalist clinicians, leaders, and administrators, as well as researchers with expertise in human factors engineering and cognitive load theory. Three rounds of surveys were conducted, during which participants provided input on the salient measures of hospitalist workload across various domains. In the first round, free-text data collected from the surveys were analyzed using a directed qualitative content approach. In the second and third rounds, participants rated each measure's relevance on a Likert scale, and consensus was evaluated using the IQR. Percentage agreement was also calculated. Seventeen individuals from 14 organizations, encompassing clinicians, leaders, administrators, and researchers, participated in 3 rounds of surveys. In round 1, participants provided 135 unique qualitative comments across 10 domains, with 192 unique measures identified. Of the 192 measures presented in the second round, 6 (3%) were considered highly relevant, and 25 (13%) were considered moderately relevant. In round 3, 161 measures not meeting consensus were evaluated, with 25 (16%) considered highly relevant and 95 (59%) considered moderately relevant. Examples of measures considered highly relevant included a patient complexity score and outcome measures such as savings from hospital days avoided and clinician turnover. In this qualitative study measuring hospitalist workload, multiple measures, including those quantifying work demands and the association of those demands with outcomes, were considered relevant for measuring and understanding workloads. The findings suggest that relying on traditional measures, such as productivity-related measures and financial measures, may offer an incomplete understanding of workloads and their association with key outcomes. By embracing a broader range of measures, organizations may be able to better capture the complexity and nuances of hospitalist work demands and their outcomes on clinicians, patients, and organizations.

A case study in applying artificial intelligence-based named entity recognition to develop an automated ophthalmic disease registry

PurposeAdvances in artificial intelligence (AI)-based named entity extraction (NER) have improved the ability to extract diagnostic entities from unstructured, narrative, free-text data in electronic health records. However, there is a lack of ready-to-use tools and workflows to encourage the use among clinicians who often lack experience and training in AI. We sought to demonstrate a case study for developing an automated registry of ophthalmic diseases accompanied by a ready-to-use low-code tool for clinicians.MethodsWe extracted deidentified electronic clinical records from a single centre’s adult outpatient ophthalmology clinic from November 2019 to May 2022. We used a low-code annotation software tool (Prodigy) to annotate diagnoses and train a bespoke spaCy NER model to extract diagnoses and create an ophthalmic disease registry.ResultsA total of 123,194 diagnostic entities were extracted from 33,455 clinical records. After decapitalisation and removal of non-alphanumeric characters, there were 5070 distinct extracted diagnostic entities. The NER model achieved a precision of 0.8157, recall of 0.8099, and F score of 0.8128.ConclusionWe presented a case study using low-code artificial intelligence-based NLP tools to produce an automated ophthalmic disease registry. The workflow created a NER model with a moderate overall ability to extract diagnoses from free-text electronic clinical records. We have produced a ready-to-use tool for clinicians to implement this low-code workflow in their institutions and encourage the uptake of artificial intelligence methods for case finding in electronic health records.

Site staff perspectives on communicating trial results to participants: Cost and feasibility results from the Show RESPECT cluster randomised, factorial, mixed-methods trial.

Sharing trial results with participants is an ethical imperative but often does not happen. Show RESPECT (ISRCTN96189403) tested ways of sharing results with participants in an ovarian cancer trial (ISRCTN10356387). Sharing results via a printed summary improved patient satisfaction. Little is known about staff experience and the costs of communicating results with participants. We report the costs of communication approaches used in Show RESPECT and the views of site staff on these approaches. We allocated 43 hospitals (sites) to share results with trial participants through one of eight intervention combinations (2 × 2 × 2 factorial; enhanced versus basic webpage, printed summary versus no printed summary, email list invitation versus no invitation). Questionnaires elicited data from staff involved in sharing results. Open- and closed-ended questions covered resources used to share results and site staff perspectives on the approaches used. Semi-structured interviews were conducted. Interview and free-text data were analysed thematically. The mean additional site costs per participant from each intervention were estimated jointly as main effects by linear regression. We received questionnaires from 68 staff from 41 sites and interviewed 11 site staff. Sites allocated to the printed summary had mean total site costs of sharing results £13.71/patient higher (95% confidence interval (CI): -3.19, 30.60; p = 0.108) than sites allocated no printed summary. Sites allocated to the enhanced webpage had mean total site costs £1.91/patient higher (95% CI: -14, 18.74; p = 0.819) than sites allocated to the basic webpage. Sites allocated to the email list had costs £2.87/patient lower (95% CI: -19.70, 13.95; p = 0.731) than sites allocated to no email list. Most of these costs were staff time for mailing information and handling patients' queries. Most site staff reported no concerns about how they had shared results (88%) and no challenges (76%). Most (83%) found it easy to answer queries from patients about the results and thought the way they were allocated to share results with participants would be an acceptable standard approach (76%), with 79% saying they would follow the same approach for future trials. There were no significant effects of the randomised interventions on these outcomes. Site staff emphasised the importance of preparing patients to receive the results, including giving opt-in/opt-out options, and the need to offer further support, particularly if the results could confuse or distress some patients. Adding a printed summary to a webpage (which significantly improved participant satisfaction) may increase costs to sites by ~£14/patient, which is modest in relation to the cost of trials. The Show RESPECT communication interventions were feasible to implement. This information could help future trials ensure they have sufficient resources to share results with participants.

Reliability ofElectronic Health Records inRecording Veterans' Tobacco Use Status.

The prevalence of tobacco use in the Veteran population and among Veterans Health Administration patients remains high, resulting in significant health and economic consequences. This problem has generated many tobacco research studies and clinical interventions, which often rely upon tobacco use status data previously recorded in electronic health records (EHR). Therefore, the consistency and reliability of these data are critical. The Veterans Health Administration uses an extensive EHR system where tobacco use status can be documented either as free text (FT) or as health factors (HF). The current literature assessing the reliability of HF and FT data is limited. This analysis evaluated the agreement between HF and FT tobacco use status data. This retrospective study included Veterans who underwent coronary revascularization and had tobacco use statuses recorded as both HF and FT. These statuses were categorized as "Current," "Former," or "Never." The closest recorded status to the index date (date of revascularization procedure) for each subject in both datasets was chosen, and Cohen's kappa statistic was calculated to measure the agreement between HF and FT. Implausible tobacco use status changes within each dataset were quantified to assess trustworthiness. Agreement between HF and FT data was first measured for all subjects (n = 1,095), which included those who had implausible status changes in either dataset and then measured again for subjects (n = 770) without any implausible status changes in either dataset. This study was exempt from institutional review board review. Overall, 14.3% and 17.7% of all subjects had implausible tobacco use status changes in HF and FT data, respectively. For all subjects (n = 1,095), including those with implausible data, there was "moderate" agreement between HF and FT data (kappa = 0.49; 95% CI, 0.44-0.53). For subjects without implausible data (n = 770), the strength of agreement between HF and FT data was "good" (kappa = 0.64; 95% CI, 0.59-0.69). Agreement between HF and FT data that document the tobacco use statuses of Veterans varied because of implausible data. HF data had fewer implausible tobacco use statuses, but FT data were recorded more frequently. Although HF and FT data can be reasonably relied upon to determine the tobacco use statuses of Veterans, researchers and clinicians must be aware of implausible data and consider methods to overcome this limitation. Future studies should investigate the ways of improving the consistency of EHR documentation by health care providers and benchmark HF and FT data against a gold standard like biochemical verification to determine accuracy.

Factors associated with injury among Maine logging workers.

Despite dramatic improvements in safety, logging remains one of the most dangerous industries in the United States. The purpose of this study was to explore longitudinal injury trends among Maine logging workers. Loggers participated in seven quarterly surveys, over the course of 18 months. Categorical and free text data related to traumatic and acute injury, musculoskeletal disorders (MSD), and chronic pain were exported from REDCap into SAS 9.4, Excel, and NVivo, for quantitative and qualitative analysis, respectively. Time to injury was modeled using two different approaches: (1) time to the occurrence of first injury modeled by proportional hazard regression and (2) an intensity model for injury frequency. Two research team members also analyzed qualitative data using a content analysis approach. During the study, 204 injuries were reported. Of the 154 participants, 93 (60.4%) reported musculoskeletal pain on at least one survey. The majority of injuries were traumatic, including fractures, sprains, and strains. Lack of health insurance was found to be related to increased risk of first injury [HR = 1.41, 95% CI = 0.97-2.04, p = 0.069]. Variables found to be related to injury intensity at the univariate level were: (1) a lack of health insurance [HR = 1.51, 95% CI = 1.04-2.20, p = 0.030], (2) age [HR for 10-year age increase;= 1.12, 95% CI = 0.99-1.27, p = 0.082], and (3) years employed in logging industry [HR for 10-year increase = 1.12, 95% CI = 0.99-1.26, p = 0.052]. Seeking medical attention for injury was not a priority for this cohort, and narratives revealed a trend for self-assessment. A variety of barriers, including finances, prevented loggers from seeking medical attention. We found that loggers still experience serious, and sometimes disabling, injuries associated with their work. It was unsurprising that many injuries were due to slips, trips, and falls, along with contact with logging equipment and trees/logs. The narratives revealed various obstacles preventing loggers from achieving optimal health. Examples included geographic distance from healthcare, lack of time to access care, and entrenched values that prioritized independence and traditional masculinity. Financial considerations were also consistently cited as a primary barrier to adequate care. There is a continued need to emphasize occupational health and safety in the logging industry. Implementation of relevant safety programs is key, but it is likely that the benefits of these will not be fully realized until a cultural shift takes place within this industry.

Education Staff and Students’ Understanding and Expectations of the Role and Qualities of the Undergraduate Nursing Clinical Facilitator

Workplace integration of undergraduate student nurses through clinical placement remains an integral part of nurse education. The role of clinical facilitators who support this integration is complex and shaped by the expectations of various key stakeholders. These multiple expectations often resulting in the role being blurred and lacking clarity. The research aim is to identify the roles and expectations of clinical facilitators from the perspectives of academics/educators and students and to explore qualities that allow clinical facilitators to conduct their role effectively. A cross-sectional survey was utilised. Exploratory factor analysis, free text data and comparative analysis were used. Findings suggest a discordance in understanding of the clinical facilitator role by educators and students. Students tended to focus on skills and knowledge without consideration of the broader nursing profession. The clinical facilitator’s ability to role model and influence is seen as a “hidden curriculum” that students are largely unaware of. The clinical facilitator role is complex, with education critical in supporting this role and thus the success of students. Illumination of the hidden nursing curriculum would enable better understanding of the clinical facilitator role in engaging students with the varying requirements of the nursing profession.

Early detection of colorectal cancer by leveraging Dutch primary care consultation notes with free text embeddings

We aimed to assess the added predictive performance that free-text Dutch consultation notes provide in detecting colorectal cancer in primary care, in comparison to currently used models. We developed, evaluated and compared three prediction models for colorectal cancer (CRC) in a large primary care database with 60,641 patients. The prediction model with both known predictive features and free-text data (with TabTxt AUROC: 0.823) performs statistically significantly better (p < 0.05) than the other two models with only tabular (as used nowadays) and text data, respectively (AUROC Tab: 0.767; Txt: 0.797). The specificity of the two models that use demographics and known CRC features (with specificity Tab: 0.321; TabTxt: 0.335) are higher than that of the model with only free-text (specificity Txt: 0.234). The Txt and, to a lesser degree, TabTxt model are well calibrated, while the Tab model shows slight underprediction at both tails. As expected with an outcome prevalence below 0.01, all models show much uncalibrated predictions in the extreme upper tail (top 1%). Free-text consultation notes show promising results to improve the predictive performance over established prediction models that only use structured features. Clinical future implications for our CRC use case include that such improvement may help lowering the number of referrals for suspected CRC to medical specialists.

Differences in public's perception of air quality and acceptability of a clean air zone: A mixed-methods cross sectional study

BackgroundAir pollution is a major cause of morbidity and mortality. Clean Air Zones (CAZs) which restrict the entry of polluting vehicles in targeted areas have been identified as potentially effective in improving health and reducing air pollution; however, their implementation can be controversial. MethodsA cross-sectional survey was completed by 1949 respondents who lived or worked in Bradford, a multi-cultural deprived city in England, between April and December 2021. Of these, 1137 were recruited from the longitudinal Born in Bradford (BiB) family cohort (families with children born in the city during 2007–2011) and 812 were from the general public. Bradford is the seventh largest metropolitan district in England and Wales with a population of over half a million mainly white British and Pakistani origin. The BiB families cohort and the general public respondents were used for descriptive analysis of perception of air quality and acceptability of CAZ, then the relationship between participants responses with demographic characteristics were investigated using the BiB families cohort. Outcomes included perceptions of air quality and acceptability of the CAZ supplemented by free-text questions. Thematic analysis was used to code free-text data. Descriptive analyses were performed on the entire sample. Latent class analysis was used to characterise participants was performed in the BiB dataset for whom detailed existing socio-demographic data were available. ResultsThe majority of participants (67%) considered improving air quality in Bradford as extremely important; 70% supported implementation of the CAZ. Three latent classes were identified within the BiB sample: deprived white British families (25%), more affluent white British families (32%) and deprived Pakistani-origin families (43%). Deprived white British (OR = 0.54, 95% CI: 0.34 to 0.84) and more affluent white British families (OR = 0.53, 95% CI: 0.36 to 0.79) were less likely to say the air quality was good/excellent when compared with deprived Pakistani-origin families. Affluent White British families were more likely to support the CAZ compared with deprived white British families (OR = 2.24; 95% CI: 1.55. to 3.25) and deprived Pakistani-origin families (OR = 2.06, 95% CI: 1.50 to 2.85). Qualitative analysis suggested that a perceived lack of cohesion in the policy and concerns about financial impacts drove negative attitudes. ConclusionFamilies in Bradford were generally supportive of the planned CAZ and efforts to reduce pollution; however, support was weaker in more deprived communities. Pakistani-origin communities living in deprived areas perceived air quality as better than other groups. Tailored approaches to communicate about the proposed benefits of policies such as CAZ prior to implementation may be an important way to increase acceptability amongst vulnerable groups.