…

The distinction between multimorbidity as a simple count of diagnoses and multimorbidity as lived experience emerged repeatedly in the study’s framing, and it carries practical consequences for how hospital services are organised. When a person arrives at an emergency department with two, three, or more chronic conditions, the presenting problem rarely maps cleanly onto a single specialty. A patient with heart failure, chronic kidney disease, diabetes, and arthritis may trigger overlapping referral pathways, each governed by its own admission criteria, monitoring protocols, and specialist thresholds. The clinicians interviewed in this study described navigating these overlapping structures in real time, often within minutes, while simultaneously managing the flow pressures that characterise acute settings. This is the practical meaning of a single-organ oriented health system: the architecture of referral, admission, and specialist expertise was largely built around discrete diseases, and patients whose conditions interact — physiologically, pharmacologically, or socially — do not fit that architecture neatly.

The Social Ecological Model, which the researchers used to organise their findings, offers a useful lens for readers unfamiliar with qualitative frameworks in health services research. Rather than locating the causes of difficult decisions inside individual clinicians’ heads, the model treats decision-making as the product of nested layers: the individual clinician’s knowledge, experience, and professional identity; the interpersonal relationships between colleagues, patients, and families; and the wider institutional and systemic environment of targets, bed capacity, staffing, and pathway design. Applying this framework to frontline decision-making revealed that no single factor dominated. A junior doctor’s confidence in managing multimorbidity, the availability of a trusted geriatrician to consult, the occupancy of the medical assessment unit, and the presence or absence of a coherent community care record all shaped the same decision, often simultaneously. This layered view helps explain why interventions aimed at only one level — for example, education alone, or pathway redesign alone — have historically produced disappointing results in multimorbidity care.

The theme of professional expertise and identity deserves particular attention because it illuminates a structural tension in modern hospital medicine. As specialties have proliferated and sub-specialisation has deepened, the generalist skills once central to hospital practice have become less visible, even as the patient population has grown steadily more complex. Participants in this study described situations where the clinician best placed to see the whole picture was not always the clinician with the clearest route to decision-making authority. The researchers’ call for enhanced generalist expertise echoes a broader international conversation about the future of general internal medicine and acute care, in which professional bodies have warned that the erosion of generalist capacity leaves health systems ill-equipped for ageing populations with multiple conditions. The study adds qualitative texture to that argument by showing how the absence of generalist confidence is experienced at the bedside: as hesitation, as defensive referral, and as reliance on heuristics to bridge the gap between what guidelines cover and what the patient actually presents with.

Heuristics — the mental shortcuts described in the second thematic category — are neither inherently good nor bad. Decision science has long recognised that experienced clinicians use pattern recognition to reach rapid, often accurate judgements under time pressure, and the study’s participants were no exception. What the interviews reveal is the specific character of heuristic reliance in multimorbidity: when no guideline addresses the interaction of five conditions and thirteen medications, clinicians fall back on familiar cases, senior opinion, and rules of thumb about risk. The researchers frame this as an adaptive strategy, a way of keeping care moving despite uncertainty, but they also record participants’ own reservations about whether such compensatory work is sustainable or sufficient. This tension between adaptation and adequacy is one of the study’s most important contributions, because it reframes the problem: the issue is not that frontline staff make poor decisions, but that the system requires so much invisible compensatory labour from them in the first place.

Relational care, the theme that bridged both categories, functions in the analysis as both a constraint and a resource. Relationships with colleagues across specialties determined whether a referral conversation could happen quickly and informally, whether a borderline admission could be discussed before it became a crisis, and whether responsibility for a complex patient could be shared rather than defensively transferred. Relationships with patients and families shaped what was known about the person behind the presentation — their goals, their treatment burden, their capacity to manage at home — information that is often decisive in multimorbidity decisions yet poorly captured in structured records. The finding that relational work both shapes what decisions are possible and provides a means of navigating uncertainty suggests that continuity of staff, not only continuity of care records, may be an underappreciated lever for improving acute care for this population. It also suggests that service models which fragment relationships — through frequent rotation, remote triage, or transactional handover — may inadvertently degrade decision quality in ways that are difficult to measure but easy for staff to describe.

The study’s methods merit a brief note for readers assessing how much weight its conclusions can bear. Forty participants, sampled purposively across roles, specialties, and regions of the UK, is a substantial sample for a qualitative interview study, and the researchers were transparent that their target was driven by diversity and data richness rather than a formal saturation test. Later interviews elaborated existing patterns rather than introducing new topics, which the authors report as a reassuring signal, though they do not claim saturation in the technical sense. The interviews were conducted by a team with no prior relationships with participants, which reduces the risk of socially desirable accounts, though it may also have limited the depth of candour that sometimes comes with established rapport. Reflexive thematic analysis, the analytic approach used, explicitly acknowledges that the researchers’ perspectives shape theme construction; the publication of peer review reports alongside the article offers readers a window into how the analysis was scrutinised.

Several limitations contextualise the findings. All participants worked within the NHS, a single publicly funded system with particular pressures — high bed occupancy, workforce shortages, and specific referral structures — so the precise character of the system pressures described may differ in other health systems, even if the underlying tension between single-condition pathways and multimorbid patients is likely to be broadly generalisable. The study captured the perspectives of staff who agreed to be interviewed, and it is possible that those most frustrated by existing structures were more motivated to participate. Interviews also capture retrospective accounts of decision-making rather than direct observation, and clinicians’ descriptions of their reasoning may differ from their reasoning in the moment. The researchers acknowledge these constraints, and the study should be read as a detailed map of how frontline staff experience and explain their decisions, not as a measurement of decision quality itself.

The study sits within the larger ADMISSION collaborative, a UK research programme funded through UK Research and Innovation and the National Institute for Health and Care Research that brings together clinical, data, social, and computational sciences with public contributors who have lived experience of multiple long-term conditions. This publication appears alongside companion papers from the same programme published on the same date, including a qualitative exploration of inpatient experiences from the patient’s perspective, indicating a deliberate effort to build a rounded evidence base from both sides of the hospital encounter. Reading the staff perspective reported here alongside patient-facing work from the same programme suggests a coherent research agenda: understanding how the mismatch between single-organ services and multimorbid patients is produced, experienced, and potentially remedied.

For policymakers and service leaders, the study’s conclusions point toward a specific set of system-level changes rather than exhortations for individual clinicians to try harder. Improved community care would reduce the information vacuum that clinicians face at the point of presentation, when the history of a person’s conditions, medications, and recent care is often scattered across unconnected records. Enhanced generalist expertise would rebuild the capacity to see whole patients rather than organ systems. Opportunities for relational decision-making — time to talk, trusted colleagues, shared forums for complex cases — would convert the interpersonal resource identified in the analysis into a designed feature of service delivery rather than an accident of staffing. Better continuity and shared ownership of care would reduce the defensive boundary-drawing that participants described when responsibility for complex patients felt dangerous to hold. Flexible care pathways would allow admission, discharge, and referral decisions to respond to the person rather than to rigid criteria built for single conditions. None of these changes is novel in isolation, but the study’s contribution is to show how they interlock: each addresses a level of the ecological model, and the analysis suggests that partial fixes at a single level are unlikely to relieve the compensatory burden that frontline staff currently carry.

Finally, the timing of this work matters. Health systems across high-income countries are confronting rising multimorbidity as populations age, and the pressures described by these UK participants — stretched acute units, fragmented community services, guidelines that do not fit — are recognisable far beyond the NHS. By documenting, in clinicians’ own words, how decisions about referrals, admissions, and care are actually made for people with multiple long-term conditions, the study provides an evidence base for redesign efforts that start from the reality of frontline practice rather than from idealised models of how decisions ought to be made. The researchers’ central message is measured but clear: adaptive strategies keep the system functioning today, but treating that adaptation as a substitute for structural change asks too much of the people the system depends on.

Subject of Research: Clinical decision-making by hospital staff for patients with multiple long-term conditions

Article Title: Front-line decision-making: a thematic analysis of interviews with hospital staff on referrals, admissions, and care for people with multiple long-term conditions

Article References: Front-line decision-making: a thematic analysis of interviews with hospital staff on referrals, admissions, and care for people with multiple long-term conditions. (n.d.). https://doi.org/10.1186/s12916-026-05236-6

Image Credits: AI Generated

DOI: 10.1186/s12916-026-05236-6

Keywords: multimorbidity, multiple long-term conditions, clinical decision-making, emergency care, hospital admissions, referrals, qualitative research, NHS, geriatric medicine, health systems, thematic analysis, relational care

The research team interviewed eleven care management staff from four geographically diverse health centers and five subject matter experts in health information technology, all drawn from OCHIN, a nonprofit that hosts a shared instance of the Epic EHR for more than 2,000 community health center clinics across 37 states, the largest such network in the country. In 2022, OCHIN activated Epic’s care management module, called Compass Rose, which is designed to support care coordination activities including the tracking of contextual drivers such as food, housing, and transportation insecurity. Yet at the time of the study, fewer than 10 percent of OCHIN member clinics had ever used the module to make and track referrals for these needs, a striking adoption gap that motivated the investigation.

Methodologically, the study was a formative evaluation informed by human-centered design principles, conducted between September 2023 and July 2024. The researchers used semi-structured interviews paired with a virtual “guided tour” technique, in which participants shared their screens and demonstrated how they actually navigated the EHR to document screening, make referrals, and follow up, using a dummy patient record. Transcripts were analyzed with a rapid qualitative analytic approach organized around the Integrated Technology Implementation Model, and the findings were distilled into a care manager journey map that traces each step of addressing contextual drivers, the tasks involved, and the barriers that surface at each one. Preliminary findings were shared back with participants to verify accuracy.

Step one of the journey, screening, is the system’s relative bright spot. Health centers screen for contextual needs at check-in, rooming, or care management enrollment, and results are captured in EHR flowsheets that support longitudinal tracking. Some centers have configured alerts that prompt any staff member opening a chart to screen, and shortcuts allow quick insertion of relevant text into notes. Once documented, social risk data appear as color-coded icons in the EHR’s patient summary and within the care management module itself, giving care managers a consolidated view. One care manager described the appeal plainly: everything from screenings to disease-specific questions is “all right there,” eliminating the need to hunt across the chart for information relevant to care planning.

Step two, making referrals, is where the machinery begins to grind. For internal referrals to social workers or community health workers, care managers can use EHR messaging and referral ordering. For external referrals to community-based organizations such as food pantries or transportation services, some centers use community resource referral systems like Findhelp or Unite Us, platforms increasingly integrated with EHRs to enable closed-loop communication with registered organizations. But adoption of these platforms is hampered by two major challenges the participants identified: the effort and cost required for both clinics and community organizations to establish the referral model, and the fact that many community-based organizations lack the technological infrastructure to engage in the required data exchange at all. Relationships, participants stressed, matter as much as software; one expert recounted how a center’s referrals went unaccepted because the organizations it wanted to reach had never been onboarded to the platform.

Documentation of referrals proved equally problematic. Most EHR tools as configured at the participating centers offer no discrete field for referral documentation outside a specific referral work queue, and that work queue demands extensive manual entry and is not linked to care management care plans. Within the care management module, care plans include only pre-set, general referral options that must be customized by hand, a task one care manager called a “huge slowdown” that is “very, very time-consuming.” Worse, care plans built inside the module do not propagate to other views of the chart, forcing care managers to re-enter the same information elsewhere so that other care team members can see it. As one staff member explained, the care plan exists only in the module and “doesn’t cross over into any other part” of the record, so visibility elsewhere requires duplicative documentation. The practical result is that much referral information ends up in free-text notes, where it cannot be systematically tracked.

Step three, tracking whether patients actually connected with services, exposes the deepest gaps. Community resource referral systems could in principle close the loop by letting community providers report service receipt back into the EHR, but their limited adoption leaves most centers without that capability. Only referrals routed through the standard EHR referral work queue can be followed longitudinally, and even then, the lack of integration with care management care plans means outcome dispositions must be re-entered in multiple locations. Referrals documented in care plans through free text or shortcuts were not reliably trackable at all. One staff member admitted that after years of doing this work, the center had no formal referral pathway to community partners and could barely “fathom how that would work” given the scale of capturing and closing referrals. Notably, the care management module contains a “tasking” dashboard feature that could support referral tracking, but care managers reported they rarely use it for this purpose simply because they never received adequate training on how.

The study’s authors argue that these barriers can be sorted into three categories: technical, organizational, and those requiring implementation support rather than tool redesign. On the technical side, structured data fields for referrals, care plans customized to each center’s specific programs and reimbursement requirements, and alignment with emerging documentation standards such as those promoted by the Gravity Project could make referral data visible, actionable, and reportable. On the organizational side, the findings highlight a structural tension: health centers are investing heavily in EHR tools and workflows, but the community organizations receiving referrals are often under-resourced to participate in electronic exchange. Without investment in the infrastructure and readiness of those partners, even well-designed platforms will be underused, a problem likely to be most acute in communities with the fewest available services. The authors suggest implementation strategies that include assessing community organization readiness, mapping feasible referral pathways, and building backup workflows when electronic exchange is not possible.

Implementation support emerges as the study’s central prescription. Participants described their training on the care management module as something they “learned as we went,” with comments like “there wasn’t much training available” and wishes for “a much better training program.” Experts confirmed that hands-on, step-by-step guidance with demonstration and practice is the single biggest lever for improving uptake, and that support must be ongoing, since staff reported questions persisting even one or two years after adopting the tools. Leadership backing, workflow redesign coaching, and site-specific configuration assessments before training begins round out the recommended support package. These findings will now feed directly into a cluster-randomized trial, registered on ClinicalTrials.gov, that will test implementation strategies aimed at helping care management teams adopt EHR-based tools for social care coordination, an effort the researchers describe as the first of its kind. The study’s limitations are acknowledged: a small sample, a single EHR system, and the absence of community organization perspectives, though the authors note that Epic’s dominance in health centers and the breadth of roles interviewed strengthen the findings’ relevance.

The broader significance is hard to overstate. Federal and state programs, from Medicare’s Chronic Care Management codes to California’s CalAIM Enhanced Care Management, increasingly require the screening of social needs and linkage to services, and value-based payment models tie reimbursement to that documentation. If the electronic backbone meant to support this work remains siloed and untracked, centers risk both missed interventions for vulnerable patients and unfulfillable reporting obligations. This evaluation offers a precise map of where the technology works, where it breaks, and what a coordinated fix would require: better configuration, shared visibility across teams, genuine two-way infrastructure with community partners, and sustained, hands-on support for the people asked to use it.

Subject of Research: Adoption of electronic health record-based care management technologies for addressing social needs in community health centers

Article Title: Community Health Center Adoption of Enabling Technologies to Address Contextual Drivers of Health for Care-Managed Patients: Formative Evaluation

Article References: Owens-Jasey, C., Gunn, R., Cook, N., Fein, H. L., Pisciotta, M., Fee, C., Larson, Z., Templeton, A., & Gold, R. (2026). Community Health Center Adoption of Enabling Technologies to Address Contextual Drivers of Health for Care-Managed Patients: Formative Evaluation. Journal of General Internal Medicine. https://doi.org/10.1007/s11606-026-10735-6

Image Credits: AI Generated

DOI: 10.1007/s11606-026-10735-6

Keywords: community health centers, electronic health records, care management, social needs, care coordination, health information technology, referrals, implementation science, contextual drivers, social determinants of health, qualitative research, health informatics