Background: Research has produced a substantial and expanding body of evidence-based practices (EBPs), encompassing interventions, programmes, clinical guidelines, protocols, care pathways and models of care supported by the best available evidence. Despite this, healthcare delivery is still frequently characterised as insufficiently evidence-based, reflecting a persistent gap between what is known to be effective and what is routinely implemented in practice. Traditional explanations only partially account for this gap, often overlooking culture as a critical yet under-theorized influence. Culture is a learned phenomenon rooted in social contexts, encompassing shared norms, values, beliefs and assumptions that define a group, whether a team, profession or organization. This paper argues that the uptake and sustainability of EBPs are profoundly shaped by cultural dynamics operating across three key layers: organizational, professional and disciplinary.

Main body: Organizational culture shapes openness to change, learning and psychological safety, influencing whether EBPs are seen as improvements or burdens. Professional cultures, rooted in education and identity, affect how physicians, nurses and other professionals apply guidelines and protocols. Disciplinary cultures, tied to clinical environments (e.g. emergency, intensive, mental health, palliative care), also shape how EBPs are received. Enhancing cultural responsiveness requires aligning EBPs with the shared norms, values, beliefs and assumptions of the intended users. Strategies include fostering clinician engagement in the development of EBPs, cultural competence, local adaptation and leveraging cultural champions.

Conclusion: Implementation of EBPs is shaped by culture, not solely by the strength of evidence or implementation strategies. Organizational, professional and disciplinary cultures interact to influence how EBPs are interpreted, accepted, adapted or resisted in practice, helping to explain persistent variation in uptake. Misalignment between EBPs and prevailing norms, values, beliefs and assumptions undermines implementation even when evidence is strong.

© The Author(s) 2026.

This critique responds to Eldh et al.'s (Implement Sci Commun 6:113, 2025) commentary on Nilsen et al.'s proposal to distinguish between implementation efficacy and effectiveness along an ideal-to-real-world continuum. While acknowledging the constructive intent of Eldh et al.'s reflections, we clarify that our framework was never intended as a simplistic, one-dimensional model but as a pragmatic heuristic to enhance design transparency. Eldh et al.'s proposed two-axis alternative is conceptually overlapping, as both axes reflect contextual variation rather than independent constructs. Our adaptation of the PRECIS framework - long validated in clinical and health services research - already incorporates multidimensional nuance through distinct domains. We emphasize that the ideal end of the continuum denotes highly supported conditions, not normative perfection. Moreover, the proposed Implementation PRECIS tool is intended to stimulate integration of contextual transparency and economic evaluation within implementation research. While we concur with Eldh et al.'s emphasis on facilitation, co-production, and contextual complexity, their critique ultimately reinforces our core premise: that explicitly positioning studies along an efficacy-effectiveness spectrum strengthens interpretability, transparency, and real-world relevance in implementation science.© The Author(s) 2026.

The present study aimed to investigate the extent to which two new implementation strategies-a new policy mandating alcohol interventions in primary care and access to online training, impacted alcohol-related clinical activities in primary care in Stockholm. This was a prospective longitudinal register-based study. One hundred twenty-nine primary care clinics in Region Stockholm agreed to provide data. The new healthcare policy was introduced in February 2021. A brief digital training for primary care professionals on managing harmful alcohol use and dependence was launched 10 months later. Seven indicators that reflect alcohol-related clinical activities were obtained from electronic case files: structured documentation on alcohol habits, the AUDIT instrument, ordering of blood tests for biomarkers of heavy drinking, prescription of medicines for alcohol dependence, registered alcohol-related diagnoses, completed advice regarding alcohol use disorder (AUD), and referrals to specialized care. Data from registers were collected before and after the policy and training was available. At baseline low levels of alcohol-related clinical activities were found in primary care. A modest, clinically non-significant increase was seen for all indicators except for frequency of prescription of medicines for alcohol dependence, over the whole follow-up. The digital training was not associated with an increase in alcohol-related clinical activities. While a policy making alcohol interventions mandatory, combined with a training program, has strong support from implementation science, only a modest, clinically non-significant increase in alcohol-related clinical activities was found. Stronger implementation strategies seem necessary to improve management of alcohol dependence in primary care. © The Author(s) 2025. Published by Oxford University Press on behalf of the European Public Health Association.

Introduction Absence of structured guidelines to navigate the complexities of implementing AI-based applications in healthcare is recognized by clinicians, healthcare leaders, and policy makers. AI implementation presents challenges beyond the technology development which necessitates standardized approaches to implementation. This study aims to explore the activities typical to implementation of AI-based systems to develop an AI implementation process framework intended to guide healthcare professionals. The Quality Implementation Framework (QIF) was considered as an initial reference framework.Methods This study employed a qualitative research design and included three components: (1) a review of 30 scientific articles describing differences empirical cases of real-world AI implementation in healthcare, (2) analysis of qualitative interviews with healthcare representatives possessing first-hand experience in planning, running, and sustaining AI implementation projects, (3) analysis of qualitative interviews with members of the research groups network and purposively sampled for their AI literacy and academic, technical or managerial leadership roles.Results The data were deductively mapped onto the steps of QIF using direct qualitative content analysis. All the phases and steps in QIF are relevant to AI implementation in healthcare, but there are specificities in the context of AI that require incorporation of additional activities and phases. To effectively support the AI implementations, the process frameworks should include a dedicated phase to implementation with specific activities that occur after planning, ensuring a smooth transition from AI's design to deployment, and a phase focused on governance and sustainability, aimed at maintaining the AI's long-term impact. The component of continuous engagement of diverse stakeholders should be incorporated throughout the lifecycle of the AI implementation.Conclusion The value of this study is the identified processual phases and activities specific and typical to AI implementations to be carried out by an adopting healthcare organization when AI systems are deployed. The study advances previous research by outlining the types of necessary comprehensive assessments and legal preparations located in the implementation planning phase. It also extends prior understanding of what the staff's training should focus on throughout different phases of implementation. Finally, the overall processual, phased structure is discussed in order to incorporate activities that lead to a successful deployment of AI systems in healthcare. © 2025 Nair, Nygren, Nilsen, Gama, Neher, Larsson and Svedberg.

Introduction: Artificial intelligence (AI) has the potential to improve the quality and efficiency of medical triage in primary care. However, there are many uncertainties related to its use. Trust in these systems is important for successful integration and advancement into healthcare, yet this remains an understudied issue. Understanding the influences on trust in the actual use of AI is necessary for developing effective implementation strategies. Objective: This study aimed to explore the influences on trust of healthcare professionals and patients in the use of AI-based triage in primary care in Sweden. Methods: We applied qualitative study design using an inductive approach based on semi-structured interviews with 14 healthcare professionals and 12 patients in two regions in Sweden. The participants had experience of using AI-based triage in primary care. The interviews were transcribed verbatim and analyzed with reflexive thematic analysis to explore the influences on trust. Results: Healthcare professionals and patients experienced three types of influences on their trust in the use of AI-based triage in primary care: (1) provision of accurate patient information, (2) alignment with clinical expertise, and (3) supervision of patients’ health and safety. Their experiences across these themes varied only in terms of the influence of experience-based knowledge. Both healthcare professionals and patients emphasized the importance of constructive dialogue, along with clear instructions for the use and storage of information. Conclusions: The results demonstrate that building trust in AI requires improved interaction to ensure that the system is adapted to the users' competencies and level of expertise. The generalizability of these insights is limited to AI-based triage in primary care in Sweden. Future research should explore trust in AI across different healthcare settings to inform policy, as well as to ensure safe use and design of AI applications. © 2025 Steerling, Svedberg, Nilsen, Siira and Nygren.

Background: Given the growing challenges in primary care, including high demand and workforce shortages, artificial intelligence (AI)-based triage applications are being explored as a means of alleviating workloads. While the potential of AI in this context is widely acknowledged, there is still limited empirical research on how such tools become embedded in routine practice, especially from healthcare professionals’ perspectives. This study focused on exploring healthcare professionals’ experiences of using an AI-based triage application in primary care.

Methods: The study had a qualitative design with a deductive approach, involving 14 healthcare professionals (physicians, nurses, psychologists, and a social worker). Data were collected through semi-structured interviews. The data were analyzed through directed qualitative content analysis and categorized in accordance with normalization process theory (NPT).

Results: The results of this study were framed by the NPT constructs: Coherence, Cognitive Participation, Collective Action, and Reflexive Monitoring. Professionals aimed to achieve Coherence by making sense of the AI triage application’s purpose and potential role in practice; however, insufficient initial information was reported to hinder a full understanding and meaningful engagement with the tool. The work of building and sustaining engagement (Cognitive Participation) was challenged by staff’s perceptions that use of the triage application was optional: this hindered the development of a “community of practice”. During Collective Action, professionals tended to rely more on patients’ free-text descriptions than on the AI-generated summaries, reflecting concerns about the application’s adequacy, compared to clinical judgment. Finally, Reflexive Monitoring revealed persistent uncertainty about the application’s value, with professionals questioning its usefulness, effectiveness, and equitable accessibility across patient groups.

Conclusions: This study found that, although the AI-based triage application appeared, at first, to be integrated into primary care practice, it was not embedded fully within professional and organizational routines. Despite a broad acceptance of digitalization among healthcare professionals, several barriers to meaningful use were identified. These included concerns about insufficient organizational and policy support, which hindered the application’s full integration into everyday workflows. The study results suggest that further efforts are needed to overcome these barriers and support the successful normalization of the AI-based triage application into routine practice. © The Author(s) 2025.

Background: The integration of artificial intelligence (AI) in radiology has advanced significantly, but research on how it affects the daily work of radiology staff is limited. Objective: This study aimed to explore the experiences of radiology staff on the integration of an AI application in a radiology department in Sweden. This understanding is essential for developing strategies to address potential challenges in AI integration and optimize the use of AI applications in radiology practice. Methods: This qualitative case study was conducted in a single radiology department with 40 employees in 1 hospital in southwestern Sweden. The study concerned the integration of an AI-powered medical imaging software designed to assist radiologists in analyzing and interpreting medical images. Using a qualitative design, interviews were conducted with 7 radiologists (physicians), 4 radiologic technologists, and 1 physician assistant. Their experience within radiology varied between 13 months and 38 years. The data were analyzed using qualitative content analysis. Results: Participants cited numerous strengths and advantages of significant value in integrating AI into radiology practice. Numerous challenges were also revealed in terms of difficulties associated with choosing, acquiring, and deploying the AI application and operational issues in radiology practice. They discussed experiences with diverse strategies to facilitate the integration of AI in radiology and to address various challenges and problems. Conclusions: The findings illustrate how AI integration was experienced in 1 hospital. While not generalizable, the study provides insights that may be useful for similar settings. Radiology staff believed AI integration enhanced decision-making and quality of care, but they encountered challenges from preadoption to routine use of AI in radiology practice. Strategies such as internal training and workflow adaptation facilitated the successful integration of AI in radiology. © Per Nilsen, Petra Svedberg, Ingrid Larsson, Lena Petersson, Jens Nygren, Emilie Steerling, Margit Neher.

Background: The distinction between efficacy (performance under ideal conditions) and effectiveness (performance in real-world settings) is well established in intervention research. Intervention effectiveness is often used as a proxy for implementation readiness. However, relying on this assumption can lead to overly optimistic expectations about real-world outcomes if the complexities of routine practice settings are not adequately considered. Main body: This paper introduces the distinction between implementation efficacy (implementation strategy performance under controlled or highly supported conditions) and implementation effectiveness (performance under typical, resource-constrained settings). We argue that the efficacy–effectiveness distinction is as critical for implementation research as it is for intervention research. Recognizing and systematically operationalizing this distinction can sharpen conceptual clarity, strengthen research design and enhance the relevance and generalizability of findings for real-world application. Yet despite its importance, this distinction is rarely made explicit in implementation studies. Research often fails to specify the conditions under which implementation strategies are investigated; studies can vary widely in how closely they reflect routine practice. Compounding this issue, economic evaluations remain uncommon in implementation research. However, without systematic assessment of resource use, it is difficult to determine whether reported implementation outcomes have been achieved through contextually feasible strategies or through intensive supports, such as dedicated staffing, external facilitation, or financial incentives, which are rarely available in everyday practice. To address this gap, we propose adapting the PRECIS-2 (Pragmatic Explanatory Continuum Indicator Summary 2) framework from clinical trials into an “Implementation PRECIS” tool. An adapted version of PRECIS-2 for implementation research could offer a systematic way to describe the extent to which a study reflects idealized conditions versus real-world practice. Conclusion: Clarifying whether implementation strategies are studied under efficacy-like or effectiveness-like conditions enhances research design, interpretation, and communication with stakeholders. It also supports informed decisions about replication and scale-up. By embracing this distinction, implementation research can temper overly optimistic assumptions, better reflect real-world constraints, and contribute more meaningfully to evidence-based practice. We argue that making this distinction explicit is a necessary step toward a more pragmatic and transparent science of implementation. © 2025 Elsevier B.V., All rights reserved.

Background: The psychosocial work environment in healthcare is widely recognized as challenging. High workload, stress, and poor work-life balance contribute to negative health outcomes for healthcare workers. Swedish primary healthcare faces similar issues, yet efforts to address them have focused primarily on individual-based interventions, such as stress management. Research on organizational initiatives remains limited, despite their greater potential for achieving long-term, sustainable improvements. Aim: This study aims to explore characteristics of primary healthcare units where organizational initiatives to improve the psychosocial work environment have been successfully carried out. Methods: A multiple case approach was used, allowing various cases to be investigated and enabling identification of similarities and common patterns across the units. Results: Four main categories and 16 subcategories were identified, capturing key factors that contribute to a favorable psychosocial work environment in primary healthcare through organizational initiatives. The main categories are engaged leadership, an open workplace climate, conditions for improvement, and a structured work organization. Conclusion: This study identifies key characteristics of primary healthcare units that contribute to creating a favorable psychosocial work environment in Swedish primary healthcare. These elements promote inclusivity, balanced change processes, and staff involvement in decision-making. The findings underscore the need for further research on managerial challenges and effective strategies for staff recruitment and retention. © 2025 Fernemark et al.

Objectives: To compare the implementation of a co-created evidence-based injury prevention intervention with an existing program in youth team handball. Design: Pragmatic two-armed cluster-randomized controlled trial, conducted collaboratively with the Swedish Handball Federation. Methods: Eighteen clubs offering handball for female and male youth players (aged 12–16 years) were allocated randomly (stratified by club size) to either intervention or control. Intervention was a holistic program integrated within handball practice (Implementing injury Prevention training ROutines in TEams and Clubs in youth Team handball) plus tailored implementation support. Control was an existing warm-up program with a few physical injury prevention principles. Implementation outcomes were investigated among coaches (both groups) and club administrators (intervention group) at the end of the handball season using a study-specific questionnaire mapped to the Reach, Effectiveness, Adoption, Implementation, and Maintenance framework. Results: No differences were observed in implementation outcomes between intervention and control coaches. Reach was high, and coaches and club administrators generally believed injury prevention training was effective. They liked the intervention and intended to use it the next season. Sixty-six percent of intervention coaches and 58 % of control coaches had used the intervention (odds ratio = 1.33, 95 % confidence interval 0.42–4.24, p = 0.625). Utilization frequency was low, and program fidelity was poor. Conclusions: A co-created holistic program supported by tailored implementation strategies did not enhance the implementation of injury prevention training compared with an existing program. Further research is needed to better understand how to enhance the implementation of such training in regular practice of youth handball. © 2025 The Authors