Electronic Clinical Decision Support System for Stroke Risk Screening in Patients With Atrial Fibrillation in Mental Health Care: Mixed Methods Study

Introduction

Electronic clinical decision support systems (eCDSSs) are software-based tools that analyze patient data locked in electronic health records (EHRs) and provide clinicians with relevant clinical support in the form of alerts or reminders [1]. Given the increasing volume of clinical information and the rapid advances in the field of medicine, eCDSSs can be pivotal in providing evidence-based clinical guidelines and tailored clinical support with personalized guidance for diagnostic, therapeutic, and preventive interventions [1].

eCDSSs have gained substantial attention in recent years for their potential to assist health care professionals in selecting appropriate treatment, managing medication (eg, dosing, contraindications, potential interactions, and side effects), calculating risk scores, identifying patients at risk, tracking patient progress over time, and documenting clinical data [2]. This has the potential to reduce medical errors and enhance health outcomes. However, eCDSSs can also have drawbacks. Alert fatigue can result in health care professionals becoming desensitized to notifications and potentially missing important information. This is often the case when the digital tool is overused or poorly designed [2-4]. Clinicians can report feeling overwhelmed with the volume and frequency of alerts, which may, in turn, disrupt workflow, resulting in less face-to-face time with patients [2-4]. Additionally, eCDSSs can lead to incorrect recommendations if the data input are inaccurate or of poor quality [2-4].

Many eCDSSs have been developed to help health care professionals manage physical health conditions, including atrial fibrillation (AF) and associated stroke risk [5]. AF is an arrhythmia characterized by irregular heartbeats. AF disrupts the ability of the heart to pump blood effectively, resulting in a higher risk of blood clotting within the left atrium of the heart and an increased risk of stroke [6]. Based on the National Institute for Health and Care Excellence (NICE) guidelines, patients with AF should undergo a stroke and bleeding risk assessment using the CHA₂DS₂-VASc and ORBIT (Outcomes Registry for Better Informed Treatment of Atrial Fibrillation) scales, respectively. NICE recommends oral anticoagulation (OAC) therapy for patients with a CHA₂DS₂-VASc score ≥2 and asks clinicians to consider anticoagulation for males with a CHA₂DS₂-VASc of 1. When the bleeding risk is low (ORBIT score <3), OAC therapy can be initiated or continued; however, when the risk is moderate or high, careful consideration of the benefits and potential risks associated with the therapy is required [7].

Research assessing the prevalence of AF among people with mental disorders is scarce. A recent nationwide population-based study reported that the risk of AF increased by 2-fold in patients with bipolar disorder or schizophrenia and by 1.5‐ to 1.7-fold in those with depression, insomnia, and anxiety disorders compared to controls [8]. Additionally, people living with a mental illness are at increased risk of cardiovascular disease (including strokes), mainly due to risk factors such as obesity, smoking, diabetes, hypertension, and dyslipidemia [8]. Despite evidence supporting the benefits of OAC therapy, people with AF and comorbid mental health conditions are less likely than the general population to be prescribed OAC therapy [9].

While many studies have evaluated the feasibility, acceptability, and effectiveness of eCDSSs in supporting the management of AF and related stroke risk in general acute hospital settings, these studies were not conducted in mental health care settings [10-19]. Implementing an eCDSS that screens for the stroke risk among patients with AF admitted to a mental health hospital is key to early prevention and quality of life improvement.

This study focuses on older adult mental health services to investigate how AF-related stroke risk is managed in individuals with comorbid severe mental illness, addressing gaps in care to improve outcomes in this high-risk population. Specifically, the study aims to explore (1) clinicians’ experiences in managing AF-related stroke prevention in secondary mental health care services and (2) their perspectives on the potential impact of an eCDSS in enhancing the quality of care in these settings.

Methods

Design

This cross-sectional study used a mixed methods research design, incorporating a short web-based survey and individual semistructured interviews.

Ethical Considerations

Ethical approval was granted by the King’s College London Research Ethics Committee, SLaM Capacity and Capability (Trust R&D Reference: R&D2023/004) and NHS (National Health Service) Health Research Authority (22/HRA/5452). The study was conducted in accordance with the principles of the Declaration of Helsinki (1996) and all applicable regulatory requirements, including but not limited to the UK policy framework for health and social care research, Trust and Research Office policies and procedures, and any subsequent amendments. Information gathered in this study was kept confidential and managed based on the Data Protection Act, NHS Caldicott Guardian, The Research Governance Framework for Health, and Social Care and HRA Approval. Informed consent was obtained from all participants before data collection. Potential participants were provided with an information leaflet outlining the study’s purpose, procedures, and their right to withdraw at any time without consequence. Participants were informed that their participation was voluntary. To ensure privacy and confidentiality, all participant data were anonymized and de-identified. Personal identifiers were removed from the transcripts, and the data were stored securely in compliance with data protection regulations. Any information that could potentially identify participants was excluded from the analysis and reporting. No compensation was provided to participants in this study, as the nature of the study did not involve any direct financial incentives for participation.

Recruitment

The study was conducted between March and May 2023 in 3 mental health of older adult (MHOA) inpatient wards at South London and NHS Foundation Trust (SLaM). These wards provide specialized care for older adults with a range of mental health conditions, often coexisting with physical health challenges. Purposive sampling was used to identify and recruit participants who were likely to provide clinical care for patients with AF and a comorbid mental health condition.

Senior management on potential wards were first approached by the research team and given brief information regarding the nature of the study and the eCDSS to be implemented. Wards that expressed an interest in the study were provided with further detailed information.

A subgroup of health care professionals, including psychiatrists and pharmacists working on recruited wards, were all invited to complete a short survey and take part in an individual interview. Potential participants varied in terms of professional seniority and clinical experience, with a focus on including diverse perspectives to enrich the findings. Potential participants were given an information leaflet and an opportunity to ask and discuss any further concerns regarding the study. If in agreement to enroll, participants were asked to sign a consent form. The number of participants required for this study was not pre-estimated and was fully dependent on theme saturation in the qualitative part.

Intervention

The eCDSS consists of a visual prompt integrated into the EHR, which is triggered whenever a patient with documented AF, either chronic or newly diagnosed upon admission, is admitted to the hospital. Using natural language processing, the system identifies references to AF in clinical notes and alerts clinicians to confirm the presence of AF, complete clinical assessments of stroke and bleeding risks using the CHA₂DS₂-VASc and ORBIT scales, and record the scores in the EHR. For patients found to have a high risk of stroke, clinicians are prompted to refer them to OAC clinics for specialized care.

Data Collection

All participants were asked to complete a short web-based survey designed to gather demographic and professional background information, including their age, gender, professional background, and years of clinical experience. The questionnaire was developed to assess clinicians’ awareness and confidence regarding AF-related stroke prevention. It included a series of statements related to their knowledge of AF guidelines, their confidence in assessing stroke and bleeding risks using the CHA₂DS₂-VASc and ORBIT scales, and their confidence in managing patients at risk of stroke. Each statement was rated on a Likert scale, ranging from strongly disagree to strongly agree (Multimedia Appendix 1). Example statements included the following: “I am confident in identifying atrial fibrillation patients eligible for oral anticoagulation therapy,” “I am confident in assessing the stroke risk using the CHA₂DS₂-VASc tool and the bleeding risk using the ORBIT tool,” and “I am confident in managing atrial fibrillation-related stroke risk in mental healthcare settings.”

The development of the questionnaire involved collaboration with field experts, including psychiatrists, general practitioners (GPs), and health care professionals with expertise in stroke prevention and mental health care. This ensured that the items were relevant to the clinical context and aligned with current guidelines for stroke prevention in patients with AF. The questionnaire was pretested with a small sample of health care professionals to ensure clarity and relevance, and minor revisions were made based on their feedback.

In addition to the survey, an interview schedule was created to explore participants’ experiences with AF-related stroke prevention in secondary mental health care services and the potential impact of an eCDSS on clinician-led care in MHOA wards. The interview topic guide was informed by feedback from field experts, ensuring that the questions were comprehensive and aligned with the research objectives (Multimedia Appendix 2).

Participants were contacted via email, and interviews were scheduled according to their availability. The interviews were conducted via Microsoft Teams, with the same researcher (DF) leading all interviews. Each interview lasted approximately 20 minutes and followed a semistructured format with key prompts to direct the discussion while allowing flexibility for participants to share their insights. All interviews were audio-recorded, transcribed verbatim, and deidentified prior to analysis.

Data Analysis

Data collected through the questionnaire were analyzed descriptively. Responses to Likert-scale items were summarized to capture the distribution of confidence levels, perceptions of current care quality, and attitudes toward the eCDSS. Demographic and professional background data were also summarized to contextualize participants’ responses.

Thematic analysis was conducted following Braun and Clarke’s framework, incorporating updated guidance from their 2023 work on good practices in thematic analysis [20]. An inductive, data-driven approach was used to allow the themes to emerge directly from the data. The analysis was conducted by 2 members of the research team (DF and HC) and involved several iterative steps. First, both researchers immersed themselves in the data by reading and rereading the transcripts to gain a comprehensive understanding of the content. Descriptive codes were independently generated for each transcript by both researchers, with codes refined and adjusted during subsequent readings. A coding framework was then collaboratively developed based on the descriptive codes, and this framework was iteratively revised to ensure alignment and accommodate different perspectives.

Codes were grouped into broader themes that reflected significant patterns in the data and addressed the research questions. This process involved exploring similarities and differences within and across transcripts and examining patterns based on participant characteristics. Themes were iteratively refined, defined, and labeled to ensure clarity, coherence, and alignment with the data. To enhance the credibility of the findings, the themes were discussed with clinical experts and refined based on their input.

Reflexivity and Methodological Rigor

The research team actively engaged in reflexivity throughout the study to address potential biases. One of the researchers (DF) brought substantial expertise in applied health informatics and the clinical implications of stroke prevention, which could have shaped their perspective during data collection and analysis. To mitigate this, the researcher frequently reflected on preconceptions and assumptions, documented potential biases, and maintained an ongoing record of decisions made during the research process. Additionally, a second researcher (HC) independently conducted parallel analyses to provide an alternative lens and challenge interpretations.

Transparency and rigor were further enhanced by explicitly acknowledging positionality and engaging clinical experts in the refinement of themes. Methodological rigor was addressed by ensuring confirmability through the maintenance of a clear audit trail of the research process. Dependability was established through the use of a systematic and replicable analytic approach. Credibility was supported by triangulation between researchers and consultation with clinical experts, while transferability was facilitated by providing a detailed description of the study context and participants to enable readers to assess the applicability of the findings to similar settings.

Results

The sample comprised 10 participants (from a total of 15 invited clinicians), of whom 6 reported their gender as female and 4 as male. Participants’ ages ranged between 25 and 46 years, with a mean age of 32 years. In terms of professional background, a slightly larger number were psychiatrists, which included 3 participants at the consultant level and 3 at a more junior level. The remaining participants (n=4) were pharmacists. The mean years of clinical experience (defined as years a health care professional has spent in clinical practice since professional qualification) was 7.25.

In total, 50% of participants (n=5) considered that AF-related stroke prevention is suboptimal on the wards where they work. Half of participants reported being confident or somewhat confident in managing AF-related stroke prevention in mental health care settings or in making referrals to OAC clinics. Around 60% reported being confident or somewhat confident using the CHA₂DS₂-VASc tool to assess the risk of stroke, whereas only 30% reported being confident or somewhat confident using the ORBIT tool to assess the risk of bleeding. Almost all participants strongly agreed that having access to an eCDSS would help them to better assess stroke and bleeding risks in patients with AF.

Thematic analysis of the interviews identified 2 overarching themes related to prevention of AF-related stroke: (1) challenges faced on wards and (2) strategies needed to improve practice (Figure 1). As for the potential impact of an eCDSS in improving quality of care, two themes emerged: (1) perceived benefits and (2) perceived risks (Figure 2).

‎

‎

Prevention of AF-Related Stroke in MHOA Wards

Challenges

Participants discussed challenges in the prevention of AF-related stroke in MHOA wards at 2 levels.

At the clinician level, many participants reported that identifying a medical history of AF from the electronic clinical notes is a challenging and time-consuming task. Some of them attributed this to poor documentation of physical health conditions in mental health care settings. Another challenge is clinician lack of knowledge and expertise in the management of physical conditions. To optimize management of physical long-term conditions, most clinicians would seek support from specialists or refer to guidelines such as NICE. Discontinuity of care provision and lack of communication between primary and secondary care were also considered obstacles in the prevention of AF-related stroke in MHOA wards. Participants expressed their concerns about the lack of coordination and follow-up with GPs and its effect on the quality of care (Figure 1 and Table 1).

At the patient level, patients with mental illness admitted to MHOA wards are complex, generally having both physical and mental health diagnoses. Illness-related symptoms (eg, delusional beliefs) and active features of illness may result in patient denial of being physically ill, saying that they want to die or refusing medication (Figure 1 and Table 1).

Strategies to Improve Practice

To improve AF-related stroke prevention in MHOA wards, most participants suggested sending alerts to clinicians on the patient EHR containing the latest guidelines, including tools for stroke and bleeding risk assessment, guidance on how to interpret the scores, and guidance on how to refer patients at high risk of stroke to OAC clinics. Although most of the information is available online, health care professionals highlighted the importance of making it easily accessible when needed to increase efficiency. They also suggested having policies at the system level for AF-related stroke management to ensure consistency and uniformity in health care provision. At a more individual level, training sessions for health care professionals on the management of AF and how to perform stroke and bleeding risk assessments based on the latest guidelines were thought helpful (Figure 1 and Table 1).

• Subject 9 (psychiatrist): “So I think like very concise and clear guidelines. And probably like a hyperlink to where you can do the CHA₂DS₂-VASc and ORBIT scoring and then maybe have the kind of action points for the outcome scores.
• Subject 4 (pharmacist): “maybe just having like policies on how to manage AF and sort of guidance. I know we have like the physical health guidelines here, but yeah, like a clear pathway would be great.”

eCDSS for the Prevention of AF-Related Stroke

Potential Benefits

Most health care professionals reported that an eCDSS for the prevention of AF-related stroke in MHOA wards would improve clinical effectiveness. This could be through spreading knowledge on the management of the condition among clinicians specialized in mental health, defining responsibilities, and ensuring consistency in decision-making. Additionally, participants emphasized the effectiveness of the tool in saving time and speeding up the clinical assessment process. They also reported that an eCDSS would be helpful in improving patient health outcomes as it will ensure faster, safer, and more comprehensive care; improve AF identification in MHOA wards; reduce the chances of getting inappropriate treatments; and ensure early stroke prevention (Figure 2 and Table 1).

• Subject 4 (pharmacist): “Prompting clinicians and also alerting them can make people feel comfortable, knowing that they’ve got, like, some sort of system in place and like everyone know where the responsibilities lie in terms of managing.”
• Subject 3 (psychiatrist): “the benefits can help you achieve something or kind of assessment risks and benefits and things a bit more quickly.”
• Subject 7 (pharmacist): “So enabling better patient care, faster, maybe more comprehensive, maybe just safer basically if it’s flagging things up.”

Potential Risks

While an eCDSS can be a very helpful tool for health care professionals, it may have potential risks; one of these is the rigidity in decision-making. Participants reported that they may become overreliant on such digital tools, which may influence their critical thinking skills. They also emphasized that errors in the accuracy of the tool may be misleading and could result in wrong recommendations. Participants were kind of worried about the increased workload caused by the digital tool and reported that annoyance and alert fatigue could be other downsides (Figure 2 and Table 1).

• Subject 3 (psychiatrist): “I think the main thing is that people can just become kind of blinkered or rigid in their decision-making and kind of forget about the specific individual factors for that patient that may be quite relevant, but don’t necessarily come up on the on the tool.”
• Subject 4 (pharmacist): “Uh harms of this system would be over reliance on electronic systems, we can become a bit over relying I think. A bit of an overreliance sometimes isn’t great.”
• Subject 6 (pharmacist): “if the electronic system has any fault to it, then they could potentially lead to a mess.”
• Subject 2 (psychiatrist): “There’s lots of things already that we have to do on ePJS and another form is likely, unless it’s really prompting, it’s likely to get forgotten and avoided actively or found to be quite annoying.”

Discussion

Principal Findings

This was an exploratory study that sought to understand mental health care professionals’ experience in the prevention of AF-related stroke and their perspective on the potential impact of an eCDSS in improving that experience. Clinicians reported many challenges related to stroke prevention in MHOA wards, including difficulty identifying patient pertinent medical history, perceived lack of knowledge and expertise in the management of physical conditions, fragmented medical care, and patient psychotic beliefs. To improve clinical practice, they suggested reminding clinicians of the latest guidelines through alerts on patient electronic records, having clear policies at the system level, and providing clinicians with training sessions on AF-related stroke management. Clinicians reported many potential benefits for the eCDSS, including improving clinical effectiveness, better identification of patients at risk, safer and more comprehensive care, consistency in decision-making, and saving time. However, they noted that the digital tool could have potential risks such as rigidity in decision-making, overreliance, reduced critical thinking, false positive recommendations, annoyance, and increased workload.

Comparison to Prior Work

Physical comorbidities among people with mental illness present complex clinical scenarios that require a specialized and holistic approach to care. Fragmentation between primary and secondary health services could contribute to uncertainty regarding which provider is responsible for the management of physical conditions among people with mental illness [21]. This could result in missed opportunities for the identification of physical conditions, which may be hampered by often poor(er) documentation in mental health services [21]. Additionally, inadequate training and lack of physical care skills may reduce mental health care professionals’ confidence in managing physical conditions [22]. Continuous training, access to resources, and specialist support are all factors that may influence the level of confidence in dealing with acute conditions considered out of their specialty [22]. Another common scenario that prevents or delays the management of physical conditions among people with mental illness is diagnostic overshadowing, which refers to the misattribution of physical symptoms to mental illness [23]. Features of the mental illness itself may also create major challenges, as people experiencing cognitive impairment, hallucinations, or delusions may not recognize or have difficulty communicating symptoms, may resist medication or struggle with medication adherence [23].

The impact of eCDSSs on AF knowledge, OAC prescription, adherence to guidelines, and patient outcomes has been investigated in general health care settings [10-19], with mixed findings on their effectiveness [10-19]. Research aiming to understand clinician perception of how an eCDSS can be supportive and useful is scarce, although this could serve as a basis for creating digital health tools that are impactful and aligned with their needs. In a study conducted in China to evaluate the acceptance of an eCDSS that automatically assesses the risks of stroke and bleeding and suggests treatments accordingly, GPs showed positive attitudes toward the digital tool, reporting that it would be helpful and would strengthen their confidence and capabilities in managing patients with AF [24]. This is consistent with results of our study, where clinicians expressed a lack of confidence in managing stroke risk related to AF and their need to refer to guidelines or to seek advice from specialists even if they already knew about current recommendations. Thus, implementing an eCDSS providing the latest guidelines, tools required to complete clinical assessments for stroke and bleeding risks, and guidance on how to interpret these scores would decrease dependence on specialist inputs and increase clinical efficiency. Our findings are also in line with those of a recent systematic review aiming to identify barriers and facilitators of using CDSSs by primary care professionals [25]. In this review, the reported benefits of the digital tools were improving quality of care, saving time, facilitating decision-making, improving professional self-confidence, and updating knowledge [25]. The main barriers were resistance or reluctance, alert fatigue, information overload, disruption of workflow, negative attitude, lack of motivation to use, lack of computer skills, and validity concerns [25].

Strengths and Limitations

This study has several strengths. First, it used both quantitative and qualitative data collection and analysis methods, which provided a comprehensive and holistic understanding of the topic of interest. While the quantitative methods offered numerical data, the qualitative approach allowed for a deeper exploration of clinician perceptions and experiences with the digital tool. Second, the study was conducted in 3 wards at South London and Maudsley NHS Foundation Trust (SLaM), which enhances the robustness, applicability, and impact of the research findings in a specific health care context. Third, 2 researchers independently worked on data extraction and analysis, which increased the rigor, transparency, and reliability of the research process. This approach also helped reduce researcher bias and validated the results.

However, the study has several limitations. First, there may have been some reluctance among health care professionals to express their lack of knowledge or confidence in assessing physical health conditions, potentially leading to reporting bias. This was mitigated by explicitly informing clinicians that the data from interviews would be deidentified and that the purpose of the study was to understand their experiences in managing AF-related stroke risk and to inform the implementation of an eCDSS in a helpful way. Second, the study focused only on psychiatrists and pharmacists, as they are typically the professionals involved in clinical assessments related to stroke and bleeding risks. Including other health care professionals with diverse clinical experiences might have enriched the findings and provided a broader perspective. Third, the sample size in this study was relatively small. However, this limitation was addressed by continuing data collection until saturation was reached, ensuring that no new themes emerged from participants’ perspectives. This approach is consistent with findings from a recent systematic review, which suggests that saturation in qualitative research can typically be achieved within 9-17 interviews [26]. Future studies could consider expanding the sample size to confirm the findings and improve the generalizability of the results.

Conclusions

The study findings indicate that adoption of an eCDSS for stroke risk screening in a psychiatric health service has the potential to be a valuable tool. However, health care organizations and clinicians need to be mindful of the challenges associated with increased workload and the potential overreliance on the system’s recommendations. To maximize the clinical benefits while minimizing the drawbacks, a balanced approach to eCDSS integration is essential. This might involve ongoing training, customization of the system to local practice, and clear guidelines on how to use eCDSS recommendations in conjunction with clinical judgment to provide patient-centered care.