Skip to main content
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2026 Nov 8;71(1):e70389. doi: 10.1111/aas.70389

Streamlining the Evidence‐to‐Practice Pathway With Computerised Decision Support in Anaesthesiology and Critical Care: A Qualitative Study of Clinician and System‐Level Interest Holder Perspectives Across Five Scandinavian Countries

Stijn Van de Velde 1,2,✉, Siri Seterelv 1,2,3, Ilkka Kunnamo 4,5, Per Vandvik 1,6, Maija‐Liisa Kalliomaki 7, Michelle Chew 8, Marius Rehn 3,9,10, Martin Ingi Sigurðsson 11, Morten Hylander Møller 12,13
PMCID: PMC13744949  PMID: 43000120

ABSTRACT

Background

Implementing evidence‐based recommendations in anaesthesiology and critical care remains challenging due to guideline, workflow and contextual barriers. Guideline‐based computerised clinical decision support systems (CDSS), embedded in structured pathways linking international guidelines to Scandinavian‐endorsed recommendations and delivered via electronic health records, may help strengthen the evidence‐to‐practice pathway. This study aimed to explore perceived barriers and facilitators to their use.

Methods

This qualitative study interviewed 10 clinicians (anaesthesiologists and intensivists; specialists and non‐specialists; < 40 and ≥ 40 years; male and female) and nine system‐level stakeholders (hospital and health authority representatives) from Denmark, Finland, Iceland, Norway and Sweden. We integrated endorsed international guidelines into a CDSS providing patient‐specific reminders linked to recommendations, as well as shared decision‐making tools and decision‐analytic components. Using a simulation application, participants explored the CDSS in a demo electronic health record. Through a framework analysis based on the GUIDES checklist, we categorised perspectives as barriers, facilitators or suggestions.

Results

Clinicians and system‐level interest holders expressed broad interest in this multifunctional CDSS to support evidence‐based, patient‐centred care and appreciated the collaborative international pathway facilitated by digital tools. Key themes included the challenges hospitals face in translating international guidelines into up‐to‐date, locally adapted guidance. Artificial intelligence (AI) was perceived as having the potential to both enhance and challenge traditional CDSS. Opportunities were identified to strengthen shared decision‐making for major decisions through the proposed system. Finally, participants highlighted the potential value of using a decision‐analytic tool to prioritise reminders, although the tool was considered conceptually challenging in its current form and would require clearer presentation before being applied in CDSS.

Conclusions

This prototype study indicates support for a multifunctional CDSS. Building on the barriers and facilitators identified in this study, interviewees advocated for real‐world evaluation with particular attention to sustainable implementation.

Editorial Comment

This study aimed to describe how ICU clinicians experience interaction in a simulated electronic health record where there were chosen elements of decision support based on international treatment guidelines incorporated into the simulation. Participants were interviewed for their perceptions of what types of decision support might be most helpful to support their clinical work in future patient information and management systems.

1. Background

Consistently applying evidence‐based clinical practice guidelines (CPGs) in critical care settings remains challenging [1]. Obstacles related to the recommendations themselves, the local environment and workflows and the complex nature of critical care can hinder the uptake of best practices [2]. Addressing these challenges is important because integrating evidence into critical care is essential for delivering optimal patient care [3].

The Scandinavian Society of Anaesthesiology and Intensive Care Medicine (SSAI) contributes to evidence‐based critical care through various initiatives, including the development and dissemination of Nordic CPGs. The endorsement of external CPGs, such as the BMJ Rapid Recommendations, may further support this effort [4, 5]. To facilitate rapid implementation of new evidence into practice, these international CPGs are developed within a digital platform that supports the evidence ecosystem [6]. The platform facilitates guideline development, publication, updating, integration into computerised clinical decision support systems (CDSS), and the generation of shared decision‐making (SDM) outputs [7]. Although CDSS have been shown to improve adherence to CPGs, they have also been criticised for lacking patient‐centredness [8, 9]. In parallel, SDM facilitates the incorporation of patient preferences, yet its implementation in critical care remains challenging despite growing recognition of its importance [10].

Against this background, linking international CPGs to Scandinavian‐endorsed recommendations in a digital platform that enables patient‐specific CDSS reminders and SDM tools delivered in electronic health records (EHR) may help address these challenges and support high‐quality, up‐to‐date, patient‐centred critical care. The aim of this study is therefore to explore perceived barriers and facilitators among frontline clinicians and system‐level stakeholders regarding this streamlined approach.

2. Methods

2.1. Study Design and Setting

This qualitative prototype study was conducted in collaboration with the SSAI Clinical Practice Committee, the MAGIC Evidence Ecosystem Foundation and Duodecim Publishing Company Ltd. The MAGIC Foundation develops BMJ Rapid Recommendations and the MAGICapp platform, while Duodecim developed Evidence‐Based Medicine electronic Decision Support (EBMEDS), a CDSS that interfaces with MAGICapp. We conducted semi‐structured interviews with participants from the five SSAI countries (Denmark, Finland, Iceland, Norway and Sweden). During the interviews, we demonstrated an interactive CDSS that we applied within a simulated EHR application. This environment allowed us to present decision‐support capabilities in realistic scenarios, helping participants visualise integration into workflows without having to use real patient data.

2.2. System Details

The EBMEDS CDSS uses patient data to provide patient‐specific reminders (Figure 1), links to guideline recommendations and decision‐support tools, including the Net Benefit Calculator (Figure 2), which supports prioritisation of interventions by weighing potential benefits and harms [11, 12, 13]. MAGICapp allows authors to create, update and disseminate structured CPGs (Figure 3) and decision aids (Figure 4) [7, 14]. To simulate integration into an EHR, we used the EBMEDS testing and Demo Application.

FIGURE 1.

FIGURE 1

Screen capture of the EBMEDS testing and demo application showing color‐coded reminders generated from patient data.

FIGURE 2.

FIGURE 2

Screen capture showing the link to the Net Benefit Calculator tool for gastro‐intestinal bleeding prevention in EBMEDS.

FIGURE 3.

FIGURE 3

Screen capture showing the link to the recommendation for gastro‐intestinal bleeding prevention in MAGICapp.

FIGURE 4.

FIGURE 4

Screen capture showing the link to the shared decision‐making tool for gastro‐intestinal bleeding prevention in MAGICapp.

Using these systems, we developed three patient cases representing realistic clinical scenarios in which CDSS were linked to SSAI‐endorsed BMJ Rapid Recommendations on gastrointestinal bleeding prevention, lumbar puncture needle type and COVID‐19 drugs [15, 16, 17]. For each case we identified EHR‐available data elements to define decision‐support triggers (Supplementary Appendix 1). For recommendations with lower‐certainty evidence, we reviewed recent systematic reviews to confirm their clinical relevance and found no new evidence that would change practice [18, 19].

2.3. Sampling

We used purposive sampling to ensure representation of diverse participant profiles and perspectives across participating countries. We aimed to interview 10 clinicians (anaesthesiologists and intensivists; specialists and non‐specialists; < 40 and ≥ 40 years; male and female) and 10 system‐level interest holders involved in hospital quality and safety or national/regional CPG and quality programmes. SSAI Clinical Practice Committee country representatives recruited the participants.

2.4. Data Collection

We developed an interview guide (Supplementary Appendix 2) organised around participant demographics, definitions of key concepts (CPGs, SDM and CDSS) and general experiences with each concept. The guide then included a demonstration of the CDSS using one of three randomly selected patient cases. Subsequent questions explored perspectives on the context, content, system and implementation of CDSS in accordance with the GUIDES checklist for implementing guidelines through CDSS [20]. The guide ended with reflections on future directions. The interviewers (SS, SV) pilot‐tested the guide to refine its clarity and flow and subsequently conducted the one‐hour semistructured interviews via Microsoft Teams. They obtained written informed consent, video‐recorded the interviews and used Microsoft Teams to generate automatic transcripts.

2.5. Data Analysis

We used a framework analysis following five steps: data familiarisation, framework identification, indexing, charting and mapping and interpretation [21]. The analysis used the GUIDES checklist as its analytical framework, with participants' perspectives categorised as barriers, facilitators or suggestions across its four domains and 16 factors [20]. One researcher (SV) coded all transcripts in ATLAS.ti using deductive codes derived from the GUIDES framework, with additional codes developed inductively from the data [22]. Participants subsequently reviewed and verified the preliminary themes [23]. We compared findings by respondent type (clinician or system‐level interest holder), CDSS functionality and case. In cases of notable differences in perspectives, we specified the respondent type, functionality or case to which they related; otherwise, we reported the findings in general following the structure of the GUIDES checklist. We classified theme intensity for both clinicians and system‐level interest holders as high (≥ 6 participants and ≥ 3 countries), medium (3–5 participants or ≥ 2 countries) or low (1–2 participants from a single country) [21]. Quotations for each theme are provided in Supplementary Appendix 3. To assess data saturation, we used code frequency analysis [24].

3. Results

3.1. Participants

Table 1 presents the characteristics of the 19 interviewees, including clinicians (n = 10), hospital representatives (n = 5) and health authority representatives (n = 4), with near‐equal representation across countries. During participant validation, 17 of 19 participants (90%) provided feedback on the preliminary themes coded by the researcher; 15 agreed with the themes and two requested minor revisions. The interviews equally covered the cases for COVID‐19 (n = 7), gastrointestinal conditions (n = 6) and lumbar puncture (n = 6).

TABLE 1.

Participant characteristics.

Clinicians n = 10 System‐level interest holders n = 9
Gender
Male 4 6
Female 6 3
Countries
Denmark 2 2
Finland 2 2
Iceland 2 1
Norway 3 2
Sweden 1 2
Age
< 40 y 5 0
> 40 y 5 9
Specialist status 6 na
Hospital representative 2 * 5
Health authorities 1 * 4

Abbreviation: Na, not applicable.

*

System‐level interest holders who also had some clinical roles.

3.2. General Perspectives Prior to CDSS Demonstration

Table 2 summarises the general perspectives on CPGs, SDM and CDSS. Many participants reported using CPGs weekly, although some felt that CPGs should be used more frequently. Among critical care patients able to engage in SDM, participants reported variation in the consistency with which clinicians applied SDM. Most participants had little or no experience with CDSS, despite the increasing use of artificial intelligence (AI) to support clinical decision‐making.

TABLE 2.

Participants' general perspectives about clinical practice guidelines (CPG), shared decision‐making (SDM) and computerised clinical decision support systems (CDSS).

Clinical practice guidelines Shared‐decision making Computerised decision support
  • CPGs are highly valued for supporting clinical decision‐making and maintaining up‐to‐date practice in time‐constrained critical care, both in and out of hospital, as well as for supporting less experienced clinicians

  • CPGs may result in a minor but acceptable reduction in clinical autonomy

  • Clinicians generally prioritise hospital guidelines, using international and national CPGs as secondary references

  • Barriers to CPGs include gaps in scope, limited consideration of patient complexity, inconsistencies between CPGs and variation in layout

  • Clinicians have difficulties staying up to date

  • CPGs not accounting for costs or environmental impact, may complicate adapting recommendations at the hospital level

  • SDM is viewed as a cornerstone of medicine and essential for decisions about surgery, ventilation, extracorporeal membrane oxygenation, dialysis, treatment withdrawal and do‐not‐resuscitate orders

  • Differences in the application of SDM may reflect variation in how SDM is interpreted, hesitancy to engage in SDM and time constraints affecting communication of complex information

  • Limited training, lack of supportive tools and patient preferences to defer decisions may form barriers to SDM

  • CDSS may help clinicians manage an expanding medical knowledge base and increasing amounts of patient‐specific information

  • CDSS may support communication about probabilities with patients and relatives

  • Participants questioned whether CDSS are sufficiently intelligent to provide reliable advice

  • Concerns about alert fatigue

3.3. Context Related Factors

Many participants viewed the CDSS as potentially reducing unwarranted variation, supporting quality improvement goals and improving clinical decision‐making. Participants differed in which CDSS functions they found most useful, with a few preferring alternative information sources such as consulting colleagues or pocket reference books.

Participants valued reminders for supporting treatment choices, prioritisation and follow‐up, but emphasised that they should not create noise or be overly authoritative. Some clinicians expressed concern about losing clinical overview, reduced independent reasoning in trainees and overreliance among busy clinicians. The recommendation function was well received, though participants noted the need to integrate hospital‐specific guidelines. Participants saw the SDM tool as helpful for preparing discussions with patients and relatives and increasing clinician confidence. They raised concerns that patients and relatives may feel overwhelmed by the outcome data, but viewed presenting practical considerations alongside outcome data as advantageous. Participants showed variable interest in the Net Benefit Calculator function, which some found relevant but unfamiliar and difficult to interpret.

Participants raised concerns about incomplete EMR data, the need for manual data entry, potential inaccuracies and the impact on CDSS performance. One noted that data protection laws restrict access to external data sources and questioned whether CDSS could help address this. Others highlighted the large volume of intensive care unit (ICU) data as a facilitator, while a few pointed to challenges in selecting appropriate values for continuously measured variables.

In acute situations, time constraints may limit CDSS use, whereas it may integrate naturally during stable phases when clinicians typically consult CPGs post‐encounter. While the CDSS may reduce the time needed to find relevant CPGs, some clinicians may instead spend additional time reviewing available content. It may also be useful in prehospital and on‐call contexts (e.g., night shifts and weekends), where staffing is lower and clinicians may rely primarily on concise features such as reminders. Sometimes hospital computers may be too slow to use the CDSS effectively. Table 3 presents the barriers and facilitators mapped to the GUIDES checklist, stratified by participant group.

TABLE 3.

Barriers and facilitators mapped to the GUIDES checklist by participant group, with data intensity reported for each theme.

graphic file with name AAS-71-0-g003.jpg

Note: ⮾ = barrier, ⊕ = facilitator, ••• = high data intensity, •• = medium data intensity, • = low data intensity, ∅ = no data present.

3.4. Content Related Factors

Participants appreciated the ability to move from recommendations to synthesised evidence and primary research, which enhanced trust. SSAI endorsement was also seen as a strength. However, some questioned aspects of the evidence synthesis (e.g., subgroups, effect size/direction) and the recommendations (e.g., risk classification), particularly the recommendation for atraumatic needles. This was challenged in relation to implementation issues such as clinician learning curves and low procedural volume, and one participant referred to adverse events associated with atraumatic needles in their hospital. Others emphasised the importance of clearly identifying the CPG panel, disclosing any AI involvement and indicating how up to date the information is, and these aspects were not consistently clear. For the Net Benefit Calculator, participants raised questions about how outcome importance values were set.

Participants emphasised that reminder relevance is essential. CDSS that do not distinguish between primary and secondary problems may mislead clinicians or generate low‐priority reminders. Relevance also depends on clinician factors, which may change over time. Some noted atraumatic needles are already standard for lumbar puncture, making reminders unnecessary, while one added that choice is largely driven by hospital availability. Participants appreciated CDSS that show which patient factors are considered and present patient data alongside recommendations, facilitating relevance checks and consideration of additional factors. Some stressed the value of being able to verify the criteria used in the decision support algorithms.

In terms of clarity about the actions suggested by the CDSS, a few participants reported that Net Benefit Calculator outputs were not straightforward to interpret, and that SDM statements (e.g., ‘x more/fewer’) are not informative without first interpreting the underlying data.

Participants highlighted that the CDSS aggregates large amounts of information, which may generate excessive reminders for multimorbid patients. Prioritisation of decision support could mitigate this. Conversely, some raised concerns that omitting information in the CDSS could lead to loss of relevant guidance and reduced user interest, and therefore emphasised the importance of presenting comprehensive advice.

3.5. System Related Factors

Participants found the interface intuitive and appreciated its multi‐layered design allowing users to choose the level of detail. The downside of the layers is substantial clicking, which many indicated should be minimised. Some noted that certain screens contained too much information and could be better structured, linking also to feedback on avoiding long lists of reminders. One participant noted that the rationale for certainty ratings in the SDM infographics is not accessible, unlike in other screens. Many preferred reminders on demand, with only the most important shown automatically or automatic reminders reserved for less experienced clinicians. One participant noted that the hospital management might want some reminders as automatic alerts. This relates to a broader discussion on functionality that allows customisation of which reminders are shown and in what mode. A few suggested organising reminders by organ system to align with clinical reasoning.

Some participants suggested extending decision support beyond physicians to nurses, pharmacists and allied health professionals. Others asked about mobile access. One hospital is exploring replacing stationary computers with mobile devices, while another has no plans to do so. A few also emphasised having decision support continuously available from ICU admission.

3.6. Implementation Related Factors

Participants from both groups emphasised securing hospital leadership support and engaging clinical champions, as younger clinicians tend to follow their lead. They stressed fostering broad clinician ownership rather than top‐down imposition. They also recommended engaging younger clinicians in testing and developing them into superusers.

In the initial phase, clinicians need to understand the CDSS to build trust and become efficient in using its components. Participants suggested that training together with continuous information supports this process. They also noted that interpreting some CDSS advice requires specific academic skills, which may need to be addressed through training.

Participants supported gradual implementation from wards to departments, and potentially to regional and national levels, focusing on high‐impact clinical areas with sufficient time between steps. This process may be slow and vulnerable to changes in strategy or project management, underscoring the need for sustained organisational structures. Maintaining CDSS credibility over time is essential, as loss of trust may be difficult to restore.

From a system perspective, participants highlighted the value of functionality that allows monitoring performance and identifying focus areas across healthcare levels.

In terms of CDSS governance, participants described diverse views ranging from ward‐level to international coordination. Some preferred higher‐level governance due to limited time among clinicians and hospital leadership, with regional structures seen as closer to end users than national ones. Participants supported national coordination for aligning EHR infrastructure and enabling learning from larger datasets, and a few suggested international coordination, noting Scandinavia's relative homogeneity. They emphasised that higher‐level governance should not exclude hospital‐level decisions on specific content. Participants also highlighted the need for thorough cybersecurity testing and legal frameworks given distributed responsibilities across clinicians and other actors. They raised costs as a barrier, citing current cost‐saving pressures and requesting cost‐effectiveness data.

Related to the pathway from international CPGs to SSAI endorsement and CDSS integration, participants found it reasonable but highlighted its sensitivity to delays across actors, with no clear solutions identified.

One participant questioned how the complexity of the overall undertaking could compete with readily available AI solutions.

3.7. Way Forward

Clinicians and system‐level interest holders from all Nordic countries expressed a desire to advance the CDSS to practical testing. Such pilots should encompass technical integration as well as user experience, including interface design, clinical utility and integration into clinical workflows. Some participants suggested introducing the concept to a broader group of interest holders, spanning from hospital settings to the governmental level. A few participants pointed to a future with decision support tools incorporating AI‐based interpretation and noted that hospitals need to prepare clinicians to use such tools effectively.

4. Discussion

Interviewees valued a streamlined process in which internationally developed CPGs are rapidly translated into actionable recommendations, endorsed by SSAI and implemented in practice via CDSS tools. These tools include patient‐specific reminders linked to recommendations, SDM tools and decision‐analytic components.

Overall, participants viewed the integrated CDSS and its enabling ecosystem as a promising concept, with implementation considered substantial but worthwhile. Taken together, this concept can be interpreted as an operationalisation of a future EBM strategy, pointing to the need for more efficient and rapid CPG production and dissemination; adoption of electronic platforms that enable continuous updating and mobile access integrated within EHRs; integration with big data sources; provision of tools to support SDM and closer integration with decision science methodologies [25].

While the concept was broadly supported, participants highlighted several key directions for future development, which we discuss further below.

4.1. Integration of AI

Our findings align with prior research showing that, despite their promise, guideline‐based CDSSs have yet to be widely adopted in routine clinical practice [20, 26]. Although the interview guide did not explicitly address AI, many participants spontaneously made the connection. AI may create new opportunities to enhance traditional CDSS; however, such systems must be trustworthy, meet international standards, undergo rigorous scientific evaluation and obtain certification prior to formal implementation [27, 28]. At the same time, readily available AI tools are increasingly used in daily practice, underscoring the need for greater efforts to develop approved, trustworthy alternatives. In the EBMEDS CDSS, the integration of AI features is under development and awaiting EU certification.

4.2. Pathway From International GL to Hospital‐Level Guidance

Strengthening this pathway requires addressing specific issues to ensure international CPGs are well‐suited for implementation. This includes considerations of practical consequences of the recommendations and attention to computability for digital integration, consistent with prior research on CPG implementability [29, 30]. However, even with high‐quality guideline development and endorsement processes, clinicians may still disagree with some recommendations. This highlights the importance of a feedback system integrated in the CDSS that identifies perceived issues and helps determine whether they stem from the CPG or from interpretation and practice. Using shared methodologies, standards and guideline platforms with structured data could further enable the transition from international CPGs to hospital‐level guidance while streamlining timely updates and local adaptation [6, 31].

4.3. Facilitating SDM in Critical Care

The demonstrated tool was perceived as useful for illustrating possible outcomes for patients and relatives, structuring discussions and increasing clinician confidence. However, clinicians also emphasised the importance of incorporating frailty‐based risk stratification as a key prognostic factor in goals‐of‐care discussions [32]. An earlier systematic review found few studies evaluating structured communication tools to support SDMs and clinicians in this process, and the evidence of their effect remains uncertain [33].

4.4. Net Benefit Calculator

The table‐format Net Benefit calculator used in this study contained many numerical outputs and was not easy to interpret. Since the completion of this study, the tool has been updated and now includes a graphical interface and plain‐language explanations of each outcome [13]. In addition, an interactive feature has been introduced that allows the relative importance of key outcomes to be discussed with the patient and adjusted in real time, with the resulting change in the net effect estimate displayed immediately.

4.5. Strengths and Limitations

A key strength of this study is the variation in treatment recommendations, including strong and conditional recommendations supported by evidence of varying quality. We included diverse clinician and system‐level perspectives on context, content, system and implementation factors, enabling a comprehensive assessment of the CDSS concept, with no notable differences between groups. This may be explained by the fact that some system representatives also worked clinically or had previously worked in clinical roles. Code frequency analysis showed that we identified 90% of the themes after a median of 8 interviews and reached saturation within a median of 16 interviews (Supplementary Appendix 4).

A limitation is that we did not include other healthcare providers or patients, which could have generated additional themes. Another limitation is that the CDSS was evaluated in a simulated environment, which limited participants' ability to fully assess real‐world workflow integration. In addition, we used an early version of the Net Benefit Calculator, which may have constrained understanding of its application. However, the study's stepwise approach allows refinement prior to real‐world testing and supports efficient use of resources, as illustrated by improvements implemented in the Net Benefit Calculator following completion of the study. Several interviewees highlighted potential facilitators, including visualising patient data alongside recommendations and enabling monitoring of guideline adherence. These features already exist in the systems, but we did not include them in the demonstration, limiting the completeness of the presentation.

Finally, the author team's subject‐matter expertise facilitated the study but may have influenced data collection and interpretation, as some authors were involved in developing the proposed system or evaluated content. To mitigate this, interviews and analysis were conducted by researchers not involved in the development of the system or its content. The research team discussed the protocol and findings in plenary meetings, and interviewees reviewed the preliminary qualitative analysis.

4.6. Implications

Participants urged moving this concept forward to pilot and implementation studies, while carefully considering the sustainability of the effort. In addition to the key directions outlined above, further areas for development include strengthening decision support for multimorbidity, consolidating fragmented patient data and improving the user interface. The study also underscores the need to address medico‐legal questions and generate cost‐effectiveness evidence to inform governance decisions. Further research should incorporate perspectives from patients, relatives and nurses.

5. Conclusions

This prototype study showed strong interest among clinicians, hospital representatives and health authorities in a multifunctional CDSS integrating patient‐specific reminders, guideline recommendations, SDM tools and decision‐analytic support to facilitate evidence‐based, patient‐centred critical care. Building on the identified barriers and facilitators, future development should focus on integrating international, national and local guidelines, advancing SDM functions, further refining decision‐analytic tools, and exploring AI‐based functions within appropriate regulatory frameworks. Interviewees emphasised the need for real‐world testing and attention to implementation sustainability.

Author Contributions

S.V.V., S.S., I.K., M.H.M., and P.V. conceived the study. S.V.V. and S.S. defined the methodology, S.V.V. wrote the protocol which all authors reviewed and commented. M.C., M.R., M.‐L.K., M.I.S., and M.H.M. recruited study participants. S.V.V. and S.S. conducted the interviews, S.V.V. was responsible for the analysis and all authors contributed to the interpretation of the findings. S.V.V. wrote the draft paper with contributions from S.S. and I.K. All authors reviewed the draft and approved the final paper. M.H.M. supervised the study. AI (ChatGPT) was used solely to improve language and flow, not for analyses or theme identification.

Funding

Funded by the South‐Eastern Norway Regional Health Authority (2021060). The funders had no role in the study or publication.

Ethics Statement

This health services study did not involve sensitive patient data and was exempt from ethical review by the Norwegian Regional Ethical Committee. The Data Protection Officer at Lovisenberg Hospital approved the data management plan, and all data handling complied with Norwegian law and GDPR.

Conflicts of Interest

P.V. is the chief scientist of MAGIC Evidence Ecosystem Foundation and a member of the board. S.V.V. and S.S. are researchers in MAGIC. I.K. is director of development at Duodecim Publishing Company Ltd./EBMEDS and a board member of Scientific Knowledge Accelerator Foundation. The other authors declare no conflicts of interest.

Supporting information

Supplementary Appendix 1: Patient case and CDSS demonstration.

Figure S1: Screen capture of the EBMEDS Testing and Demo Application showing color‐coded reminders generated from patient data.

Figure S2: Screen capture showing the link to the recommendation for systemic corticosteroids for patients with severe or critical COVID‐19 in MAGICapp.

Figure S3: Screen capture showing the link to the shared decision‐making tool for systemic corticosteroids for patients with severe or critical COVID‐19 in MAGICapp.

Figure S4: Screen capture showing the link to the Net Benefit Calculator tool for systemic corticosteroids versus standard care in EBMEDS.

Figure S5: Improved version of the Net Benefit Calculator interface designed to reduce information density in EBMEDS.

Figure S6: Screen capture showing the link to the recommendation for gastro‐intestinal bleeding prevention in MAGICapp.

Figure S7: Screen capture showing the link to the shared decision‐making tool for gastro‐intestinal bleeding prevention in MAGICapp.

Figure S8: Screen capture showing the link to the Net Benefit Calculator tool for gastro‐intestinal bleeding prevention in EBMEDS.

Figure S9: Screen capture showing the link to the recommendation for atraumatic versus conventional lumbar puncture needles in MAGICapp.

Figure S10: Screen capture showing the link to the shared decision‐making tool for atraumatic versus conventional lumbar puncture needles in MAGICapp.

Figure S11: Screen capture showing the link to the Net Benefit Calculator tool for atraumatic needle versus conventional needle in lumbar puncture in EBMEDS.

Supplementary Appendix 2: Interview guide.

Supplementary Appendix 3: Quotes illustrating the themes.

Supplementary Appendix 4: Data saturation.

Figure S12: Results of the code frequency analysis where each line (n = 10) represents the same set of interviews in a different random order.

AAS-71-0-s001.docx (2.5MB, docx)

Acknowledgements

We thank the clinicians and interest holders from hospitals and health authorities for contributing their perspectives and insights to the interviews and we gratefully acknowledge the South‐Eastern Norway Regional Health Authority for their financial support.

Data Availability Statement

Ethical and privacy concerns prevent us from sharing raw interview data; however, we provide quotations from the transcripts in Supplementary Appendix 3 to support the findings.

References

  • 1. Weiss C. H., “Why Do We Fail to Deliver Evidence‐Based Practice in Critical Care Medicine?,” Current Opinion in Critical Care 23 (2017): 400–405, 10.1097/MCC.0000000000000436. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Rosa R. G., Teixeira C., and Sjoding M., “Novel Approaches to Facilitate the Implementation of Guidelines in the ICU,” Journal of Critical Care 60 (2020): 1–5, 10.1016/j.jcrc.2020.07.014. [DOI] [PubMed] [Google Scholar]
  • 3. Søreide E., Kalman S., Åneman A., Nørregaard O., Pere P., and Mellin‐Olsen J., “Shaping the Future of Scandinavian Anaesthesiology: A Position Paper by the SSAI,” Acta Anaesthesiologica Scandinavica 54 (2010): 1062–1070. [DOI] [PubMed] [Google Scholar]
  • 4. Siemieniuk R. A., Agoritsas T., Macdonald H., Guyatt G. H., Brandt L., and Vandvik P. O., “Introduction to BMJ Rapid Recommendations,” BMJ 354 (2016): i5191, 10.1136/bmj.i5191. [DOI] [PubMed] [Google Scholar]
  • 5. Rehn M., Chew M. S., Olkkola K. T., Örn Sverrison K., Yli‐Hankala A., and Møller M. H., “Endorsement of Clinical Practice Guidelines by the Scandinavian Society of Anaesthesiology and Intensive Care Medicine,” Acta Anaesthesiologica Scandinavica 63 (2019): 161–163, 10.1111/aas.13306. [DOI] [PubMed] [Google Scholar]
  • 6. Vandvik P. O. and Brandt L., “Future of Evidence Ecosystem Series: Evidence Ecosystems and Learning Health Systems: Why Bother?,” Journal of Clinical Epidemiology 123 (2020): 166–170, 10.1016/j.jclinepi.2020.02.008. [DOI] [PubMed] [Google Scholar]
  • 7. Vandvik P. O., Brandt L., Alonso‐Coello P., et al., “Creating Clinical Practice Guidelines We Can Trust, Use, and Share: A New Era Is Imminent,” Chest 144 (2013): 381–389. [DOI] [PubMed] [Google Scholar]
  • 8. Kwan J. L., Lo L., Ferguson J., et al., “Computerised Clinical Decision Support Systems and Absolute Improvements in Care: Meta‐Analysis of Controlled Clinical Trials,” BMJ 370 (2020): m3216, 10.1136/bmj.m3216. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Rajput V. K., Dowie J., and Kaltoft M. K., “Are Clinical Decision Support Systems Compatible With Patient‐Centred Care?,” Studies in Health Technology and Informatics 270 (2020): 532–536, 10.3233/SHTI200217. [DOI] [PubMed] [Google Scholar]
  • 10. Wubben N., van den Boogaard M., van der Hoeven J. G., and Zegers M., “Shared Decision‐Making in the ICU From the Perspective of Physicians, Nurses and Patients: A Qualitative Interview Study,” BMJ Open 11 (2021): e050134, 10.1136/bmjopen-2021-050134. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Alper B. S., Oettgen P., Kunnamo I., et al., “Defining Certainty of Net Benefit: A GRADE Concept Paper,” BMJ Open 9 (2019): e027445, 10.1136/bmjopen-2018-027445. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Djulbegovic B., Hozo I., Kunnamo I., and Guyatt G., “Improving Guideline Development Processes: Integrating Evidence Estimation and Decision‐Analytical Frameworks,” Journal of Evaluation in Clinical Practice 31 (2025): e70051, 10.1111/jep.70051. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Computable Publishing , The Net Benefit Calculator [Internet] (Computable Publishing, 2026). [Google Scholar]
  • 14. Heen A. F., Vandvik P. O., Brandt L., et al., “Decision Aids Linked to Evidence Summaries and Clinical Practice Guidelines: Results From User‐Testing in Clinical Encounters,” BMC Medical Informatics and Decision Making 21 (2021): 202, 10.1186/s12911-021-01541-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Sverrisson K. Ö., Chew M. S., Olkkola K. T., Rehn M., Yli‐Hankala A., and Møller M. H., “Clinical Practice Guideline on Gastrointestinal Bleeding Prophylaxis for Critically Ill Patients: Endorsement by the Scandinavian Society of Anaesthesiology and Intensive Care Medicine,” Acta Anaesthesiologica Scandinavica 65 (2021): 549–550, 10.1111/aas.13773. [DOI] [PubMed] [Google Scholar]
  • 16. Møller M. H., Chew M. S., Olkkola K. T., Rehn M., Yli‐Hankala A., and Sigurðsson M. I., “Therapeutics and COVID‐19‐A Living WHO Guideline: Endorsement by the Scandinavian Society of Anaesthesiology and Intensive Care Medicine,” Acta Anaesthesiologica Scandinavica 66 (2022): 636–637, 10.1111/aas.14046. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Rehn M., Chew M. S., Olkkola K. T., Sverrison K. Ö., Yli‐Hankala A., and Møller M. H., “Clinical Practice Guideline on Atraumatic (Pencil‐Point) vs. Conventional Needles for Lumbar Puncture: Endorsement by the Scandinavian Society of Anaesthesiology and Intensive Care Medicine,” Acta Anaesthesiologica Scandinavica 63 (2019): 438–439, 10.1111/aas.13312. [DOI] [PubMed] [Google Scholar]
  • 18. Wang Y., Parpia S., Ge L., et al., “Proton‐Pump Inhibitors to Prevent Gastrointestinal Bleeding—An Updated Meta‐Analysis,” NEJM Evidence 3 (2024): EVIDoa2400134. [DOI] [PubMed] [Google Scholar]
  • 19. COVID‐19 Living Network Meta‐Analysis (2021) [Internet], https://www.covid19lnma.com/. [Google Scholar]
  • 20. Van de Velde S., Kunnamo I., Roshanov P., et al., “The GUIDES Checklist: Development of a Tool to Improve the Successful Use of Guideline‐Based Computerised Clinical Decision Support,” Implementation Science 13 (2018): 86. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Goldsmith L. J., “Using Framework Analysis in Applied Qualitative Research,” Qualitative Report 26 (2021): 2061–2076. [Google Scholar]
  • 22. Gale N. K., Heath G., Cameron E., Rashid S., and Redwood S., “Using the Framework Method for the Analysis of Qualitative Data in Multi‐Disciplinary Health Research,” BMC Medical Research Methodology 13 (2013): 117, 10.1186/1471-2288-13-117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Creswell J. W. and Miller D. L., “Determining Validity in Qualitative Inquiry,” Theory Into Practice 39 (2000): 124–130, 10.1207/s15430421tip3903_2. [DOI] [Google Scholar]
  • 24. Hennink M. and Kaiser B. N., “Sample Sizes for Saturation in Qualitative Research: A Systematic Review of Empirical Tests,” Social Science & Medicine 292 (2022): 114523. [DOI] [PubMed] [Google Scholar]
  • 25. Djulbegovic B. and Guyatt G. H., “Progress in Evidence‐Based Medicine: A Quarter Century on,” Lancet 390 (2017): 415–423, 10.1016/S0140-6736(16)31592-6. [DOI] [PubMed] [Google Scholar]
  • 26. Delourme S., Redjdal A., Bouaud J., and Seroussi B., “Leveraging Guideline‐Based Clinical Decision Support Systems With Large Language Models: A Case Study With Breast Cancer,” Methods of Information in Medicine 63 (2024): 85–96, 10.1055/a-2528-4299. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Hoffmann M., Vander Stichele R., Bates D. W., et al., “Guiding Principles for the Use of Knowledge Bases and Real‐World Data in Clinical Decision Support Systems: Report by an International Expert Workshop at Karolinska Institutet,” Expert Review of Clinical Pharmacology 13 (2020): 925–934, 10.1080/17512433.2020.1805314. [DOI] [PubMed] [Google Scholar]
  • 28. Sousa‐Pinto B., Marques‐Cruz M., Neumann I., et al., “Guidelines International Network: Principles for Use of Artificial Intelligence in the Health Guideline Enterprise,” Annals of Internal Medicine 178 (2025): 408–415, 10.7326/annals-24-02338. [DOI] [PubMed] [Google Scholar]
  • 29. Seterelv S. S., Gupta S., Kouri A., et al., “From International Guidelines to Practice: A Multimethod Evaluation of BMJ Rapid Recommendations Focusing on Suitability for Adaptation and Implementation,” BMJ Open 15 (2025): e114609, 10.1136/bmjopen-2025-114609. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Johansson M., Guyatt G., and Montori V., “Guidelines Should Consider Clinicians' Time Needed to Treat,” BMJ 380 (2023): e072953, 10.1136/bmj-2022-072953. [DOI] [PubMed] [Google Scholar]
  • 31. Zaror C., Klenner M., Song Y., et al., “Adapting World Health Organization COVID‐19 Living Guidelines Balancing Methodological Rigor With Efficiency and Flexibility: A Case Study From Argentina,” Journal of Clinical Epidemiology 184 (2025): 111823, 10.1016/j.jclinepi.2025.111823. [DOI] [PubMed] [Google Scholar]
  • 32. Mowbray F. I., Manlongat D., Correia R. H., et al., “Prognostic Association of Frailty With Post‐Arrest Outcomes Following Cardiac Arrest: A Systematic Review and Meta‐Analysis,” Resuscitation 167 (2021): 242–250, 10.1016/j.resuscitation.2021.06.009. [DOI] [PubMed] [Google Scholar]
  • 33. Oczkowski S. J. W., Chung H.‐O., Hanvey L., Mbuagbaw L., and You J. J., “Communication Tools for End‐Of‐Life Decision‐Making in the Intensive Care Unit: A Systematic Review and Meta‐Analysis,” Critical Care (London, England) 20 (2016): 97. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Appendix 1: Patient case and CDSS demonstration.

Figure S1: Screen capture of the EBMEDS Testing and Demo Application showing color‐coded reminders generated from patient data.

Figure S2: Screen capture showing the link to the recommendation for systemic corticosteroids for patients with severe or critical COVID‐19 in MAGICapp.

Figure S3: Screen capture showing the link to the shared decision‐making tool for systemic corticosteroids for patients with severe or critical COVID‐19 in MAGICapp.

Figure S4: Screen capture showing the link to the Net Benefit Calculator tool for systemic corticosteroids versus standard care in EBMEDS.

Figure S5: Improved version of the Net Benefit Calculator interface designed to reduce information density in EBMEDS.

Figure S6: Screen capture showing the link to the recommendation for gastro‐intestinal bleeding prevention in MAGICapp.

Figure S7: Screen capture showing the link to the shared decision‐making tool for gastro‐intestinal bleeding prevention in MAGICapp.

Figure S8: Screen capture showing the link to the Net Benefit Calculator tool for gastro‐intestinal bleeding prevention in EBMEDS.

Figure S9: Screen capture showing the link to the recommendation for atraumatic versus conventional lumbar puncture needles in MAGICapp.

Figure S10: Screen capture showing the link to the shared decision‐making tool for atraumatic versus conventional lumbar puncture needles in MAGICapp.

Figure S11: Screen capture showing the link to the Net Benefit Calculator tool for atraumatic needle versus conventional needle in lumbar puncture in EBMEDS.

Supplementary Appendix 2: Interview guide.

Supplementary Appendix 3: Quotes illustrating the themes.

Supplementary Appendix 4: Data saturation.

Figure S12: Results of the code frequency analysis where each line (n = 10) represents the same set of interviews in a different random order.

AAS-71-0-s001.docx (2.5MB, docx)

Data Availability Statement

Ethical and privacy concerns prevent us from sharing raw interview data; however, we provide quotations from the transcripts in Supplementary Appendix 3 to support the findings.


Articles from Acta Anaesthesiologica Scandinavica are provided here courtesy of Wiley

RESOURCES