Study design

This study is a sequential exploratory mixed-methods study aimed at developing and assessing the feasibility of a provider-level multimorbidity care model [10,11] comprising three interrelated phases: (1) Evidence synthesis to inform model development; (2) evidence-driven model construction; and (3) expert feasibility consultation. Integration occurred progressively across phases: review-derived domains formed the structural foundation of the model, model construction formalised the component architecture, and expert consultation assessed feasibility and adaptability in the target implementation context. The objective was model development and feasibility assessment, not effectiveness testing.

Phase 1: Evidence synthesis to inform model development

To inform model development, we conducted a structured evidence-synthesis phase aimed at identifying operational domains, implementation approaches, and recurrent gaps in multimorbidity care models relevant to service delivery redesign in fragmented health systems [12,13]. This phase was designed to support conceptual model construction rather than to produce a standalone scoping or systematic review.

Searches were conducted between May and October 2024 in PubMed/MEDLINE, Scopus, Web of Science, and Google Scholar, covering publications from January 2000 to October 2024 in English and Spanish. An updated search was undertaken in November–December 2025 to identify additional relevant publications through that date. Structured grey-literature searches were also performed in repositories of the WHO, World Bank, PAHO, and the Colombian Ministry of Health. Colombian World Bank institutional reports were included as contextual evidence sources given their direct relevance to the study setting and research objective [8,9,14,15].

Titles, abstracts, and full texts were screened against predefined eligibility criteria (S1 Fig) by two researchers independently; eligibility decisions were documented and discrepancies resolved through discussion. Screening decisions were documented and resolved through iterative review against the study protocol. Data extraction was performed by the primary author using a structured template capturing governance arrangements, stratification mechanisms, care coordination processes, team configuration, information systems, and evaluation focus; extracted data were reviewed in full by a second researcher to verify completeness and consistency. Records were managed in Rayyan.

The synthesis used a framework-mapping approach [12]. Extracted elements were coded deductively against established integrated care frameworks, including the WHO Integrated People-Centred Health Services framework, the JA-CHRODIS Integrated Multimorbidity Care Model, and the Valentijn integrated care typology [5,6,16], while allowing inductive identification of context-specific themes emerging from the included sources. Coding outputs were compared iteratively, and discrepancies were resolved through protocol-guided review. This process generated a final set of structural domains and cross-cutting implementation gaps that directly informed model construction. Domains were further refined through cross-model comparison of European, Latin American, and Colombian sources. Of the 53 sources meeting eligibility criteria, all contributed to domain identification and gap characterisation through framework-mapping. Seventeen evidence entries — drawn from 14 distinct source references with direct, prospectively determined traceability to specific IPAM component design decisions — are presented in the traceability matrix (S1 Table). The criterion for inclusion in S1 Table was whether a source informed a specific component specification, rather than contributing solely to a thematic domain characterisation or a corroborating finding. The remaining sources informed the identification of structural domains and cross-cutting implementation gaps and are cited in the main text and reference list accordingly.

Because the purpose of this evidence-synthesis phase was to identify design-relevant domains and implementation gaps to support model development, rather than to estimate effect sizes or formally assess intervention effectiveness, a formal critical appraisal of individual sources was not undertaken [13].

Phase 2: Model construction

Model construction translated evidence-derived domains into a conceptual provider-level care framework using abductive logic – iterative movement between evidence, contextual findings, and theory to generate and refine design propositions [10]. The Consolidated Framework for Implementation Research (CFIR) [17] informed identification of modifiable inner-setting constraints relevant to IPS-level implementation. Scoping-derived domains and cross-model gaps were mapped to IPS-modifiable processes and organised into model components based on their functional purpose, responsible actors, and minimum expected process outputs. The resulting component specification informed subsequent expert feasibility consultation.

The Integrated Provider-level Adaptive Multimorbidity Model (IPAM) comprises five interdependent components. The model does not propose reform of financing or insurance architecture; rather, it positions the IPS as an operational first-mover unit by reorganising internal processes, clarifying roles, formalising coordination, and embedding proactive follow-up. Cross-level communication with other providers, insurers, and community networks is defined as a medium-term objective, contingent on the maturation of governance capacity and enabling technology.

Phase 3: Expert feasibility consultation

The preliminary IPAM underwent structured feasibility consultation through an in-person workshop involving 11 stakeholders: (1) general practitioner and (1) nurse with multimorbidity experience, (2) public health professionals, (3) health administrators, (2) Health services Quality experts and (2) EPS representatives from both contributory and subsidised insurance regimes. The workshop was framed as an expert consultation rather than a formal validation exercise, consistent with the purposive sample size and the developmental stage of the model and was conducted in April 2025.

The workshop incorporated structured elements of the Nominal Group Technique (NGT) [18,19] to support balanced participation and transparent prioritisation. Participants independently reviewed IPAM components and assessed their applicability within the current EPS–IPS context, drawing on evidence-synthesised implementation gaps and their professional experience across clinical, administrative, and insurer roles. A facilitated discussion elicited feasibility considerations, contextual constraints, and suggested refinements for each component. Priorities were agreed through structured group deliberation across two NGT cycles. Proceedings were captured in structured minutes; synthesised feedback informed component refinements without altering the five-component architecture.

Component-level ratings were collected on a five-point Likert scale across three dimensions: operational feasibility, clinical relevance, and territorial adaptability. Mean scores and ranges were calculated per component and overall; interquartile range was reported selectively where score dispersion exceeded the modal response, providing additional distributional information without assuming normality. Modifications requested during deliberation were documented and mapped to the corresponding component.

Ethical considerations

This study did not involve primary data collection from human participants beyond expert stakeholders who participated as professional informants in a service design consultation. This study was approved by the Ethics Committee of Fundación Santa Fe de Bogotá (Communication 16596, June 2024). The expert consultation was conducted with the informed agreement of all participants and informed consent was obtained through signed attendance records, which documented their voluntary agreement to participate and to allow the use of anonymised information generated during the consultation. No identifiable personal data were collected or retained. The consultation did not involve patients. All sources used in the evidence-synthesis phase were publicly available.

Reporting

The overall study is reported as a sequential exploratory mixed-methods study. The evidence-synthesis component is described in sufficient detail to ensure transparency and reproducibility, the expert consultation is reported following NGT reporting conventions [18,19].

Findings from the evidence-synthesis phase

The final sample comprised 53 sources: 29 peer-reviewed publications (including three identified through the updated search), 16 grey literature documents, and eight institutional reports (S1 Fig). The evidence synthesis yielded 53 sources meeting eligibility criteria across five structural domains: (1) population-based risk stratification; (2) care delivery redesign with multidimensional assessment and medication reconciliation; (3) governance and communication mechanisms; (4) proactive follow-up and self-management support; and (5) enabling technology with graduated implementation. Four cross-cutting implementation gaps were consistently identified: absence of multimorbidity-specific stratification tools adapted for IPS-level use; lack of standardised individualised care plan (ICP) documentation; weak provider-level governance infrastructure; and technology heterogeneity incompatible with high-entry-point requirements. Table 1 presents the key models identified and their principal characteristics. Additional evidence on care model elements and implementation gaps in LMIC settings was provided by two targeted systematic reviews included in the updated search [7,33,34].

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Key multimorbidity care models identified in the integrated scoping review.

https://doi.org/10.1371/journal.pone.0355752.t001

IPAM framework

Evidence synthesis informed construction of the IPAM. The model is organised around five interdependent components within three domains: (1) population-based management and risk stratification; (2) clinical coordination and continuity; and (3) governance and system integration at provider level. Territorial adaptability is a cross-cutting design criterion. The IPAM is designed as a sequential first step: IPS-level process standardisation (C1–C3) creates the organisational foundation upon which proactive follow-up (C4) and cross-level communication enabled by technology (C5) become operationally achievable. This sequencing responds directly to the World Bank pilot’s finding that coordination mechanisms not rooted in institutional process redesign cannot be sustained [9,15]. Table 2 maps each component to its evidence base and the specific gap addressed.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. IPAM components: evidence base, gaps addressed, and World Bank Colombia pilot findings.

https://doi.org/10.1371/journal.pone.0355752.t002

Component 1: Population-based health management (C1)

C1 establishes systematic identification and risk stratification of multimorbidity patients within IPS. Unlike MACEP’s centralised ACG-dependent analytics [25–27] or the World Bank pilot’s EPS-level identification [9], the IPAM defines minimum stratification criteria adaptable to local data capacity: multimorbidity burden, functional status, and frailty assessment for adults ≥65.

The selection of these three criteria reflects a convergence of epidemiological evidence, evaluated model design, and contextual feasibility constraints [39]. Multimorbidity burden — operationalised as the number of concurrent chronic conditions — is the most consistently documented predictor of high healthcare utilisation, care complexity, and adverse outcomes across LMIC settings [1,3,7]; Oliveira et al. confirm that condition count remains the most operationalisable stratification criterion in primary care settings with limited data infrastructure [34]. Functional status is included because disability and functional decline are the primary mediators between multimorbidity burden and high-cost care trajectories [3], and because functional assessment does not require centralised databases or validated population-level tools — both of which are unavailable across a substantial proportion of Colombian IPS [9,34]; evaluated models including JA-CHRODIS and the Cochrane synthesis identify functional assessment as a key stratification variable and moderator of intervention effectiveness [6,20,35]. Frailty assessment for adults ≥65 is included as a third criterion because the co-occurrence of frailty and multimorbidity amplifies risk of functional dependence, unplanned hospitalisation, and care fragmentation disproportionately in this age group [3]; its inclusion as a mandatory stratification component is explicitly recommended in the Italian multimorbidity guidelines [22] and operationalised in the JA-CHRODIS implementation framework [20]. Taken together, these three criteria identify patients at highest risk of high-need, high-cost trajectories while remaining implementable with paper-based registries at IPS entry level — a design constraint directly informed by the technology barriers documented in the World Bank pilot [9] and the centralised-database dependency that limits the transferability of ACG-based tools to fragmented LMIC contexts [25,26,34].

Geolocation and territorial mapping enable prioritisation of outreach for dispersed or socially vulnerable populations, responding to the geographic disparities documented in the national assessment [8]. This component does not require IT infrastructure beyond a structured patient registry, which may be paper-based at entry level.

Component 2: Care delivery redesign (C2)

C2 formalises clinical process redesign within IPS. It mandates structured multidimensional assessment, systematic medication reconciliation with defined operational steps, and documented individualised care plans (ICP) co-designed with patients using shared decision-making. Evidence synthesis and gap analysis identified conceptual ambiguity in reconciliation and absent ICP documentation as consistent cross-model gaps [7,33]; the World Bank pilot identified the same as direct barriers to care quality [9]. C2 specifies reconciliation as a distinct clinical procedure — documented in the health record — and the ICP as a mandatory output of initial multimorbidity assessment. Interdisciplinary roles are clarified, consultation workflows reorganised for comprehensive assessment, and structured case conferences institutionalised for complex cases.

Component 3: Governance and communication (C3)

C3 addresses the governance gap exposed by the World Bank pilot: when coordination depends on a single case manager role rather than institutional process design, sustainability collapses when dedicated staffing is withdrawn [9,15]. C3 formalises provider-level governance through: interprofessional case conferences (minimum monthly), shared care documentation with all specialist referrals, direct patient–institution communication channels, an institutional multimorbidity committee with assigned resources, and performance monitoring using the Table 3 indicator matrix. Communication pathways with other levels of care are defined as structured objectives, building progressively as institutional capacity matures. The evidence from high-performing IPS in the pilot [15] — where protected MDT time, systematic registries, and direct patient communication were the distinguishing features — is directly operationalised here as minimum standards.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. IPAM performance indicator framework (structure–process–outcome logic; Donabedian [40]).

https://doi.org/10.1371/journal.pone.0355752.t003

Component 4: Proactive follow-up and self-management support (C4)

C4 reconceptualises follow-up as proactive, continuous, and bidirectional. Scheduled outreach, adherence monitoring, deprescribing review, symptom reporting channels, and structured psychosocial and caregiver support are institutionalised within IPS workflows rather than delegated to a dedicated external role. The World Bank pilot demonstrated that caregiver integration and proactive contact were the most valued elements by participants [15]; the MACEP evaluation confirmed proactive contact rate as a predictor of effectiveness [29]. Operational safeguards include verified contact systems and patient registries to prevent loss to follow-up.

Component 5: Enabling technology (C5)

C5 is explicitly transversal — a cross-cutting enabler that progressively unlocks the coordination potential of C1–C4 rather than a prerequisite that must be satisfied before implementation begins. This component responds directly to the World Bank pilot finding that technology heterogeneity limited participation of lower-capacity IPS [9]. Implementation is graduated across three phases: Phase 1 (minimum viable) accepts paper-based systems; Phase 2 adds EHR templates, telehealth capacity, and drug interaction alerts; Phase 3 includes analytics dashboards and interoperable cross-level records.

Expert feasibility consultation

Score distributions indicated high consensus across most components and dimensions. C1 and C2 showed the narrowest score ranges (4–5 across all dimensions), reflecting near-unanimous agreement on the operational feasibility and clinical relevance of population-based stratification and care delivery redesign within the IPS context. C3 and C4 operational feasibility scores showed slightly greater dispersion (range 3–5; IQR 4–5 and 4–4, respectively), indicating that while most participants rated these components favourably, a minority expressed moderate reservations regarding governance formalisation and the operational demands of institutionalised proactive follow-up. C5 showed the greatest score dispersion on operational feasibility (range 2–5; IQR 3–4; mean 3.8/5), consistent with participants’ recognition that technology capacity varies substantially across IPS settings; this pattern directly reinforces the rationale for a graduated rather than prerequisite technology implementation approach. Table 4 presents component-level results and score distributions.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Expert feasibility consultation results by IPAM component (n = 11 stakeholders; Nominal Group Technique).

https://doi.org/10.1371/journal.pone.0355752.t004

Three cross-component implementation priorities emerged from NGT deliberation. First, EPS contracting incentive alignment was identified by all participants as the single most critical sustainability enabler, directly consistent with the World Bank pilot’s finding that dedicated case manager financing was unsustainable without explicit EPS commitment [9,15] and with international evidence that value-based payment models require structured provider accountability mechanisms to sustain coordination [37]. Second, protected interdisciplinary team time, currently absent from standard IPS scheduling was unanimously identified as a structural prerequisite for operationalising case conferences and multidimensional assessment workflows. Third, incorporation of social determinants — particularly housing and food insecurity — into C1 stratification criteria was prioritised in response to disparities documented in the national assessment [8].

No component was rated below 3.8/5 on any dimension, and no participant recommended removing any component or substantially altering the five-component architecture.

The traceability matrix mapping each evidence source and implementation gap to the corresponding IPAM design response is provided in Supplementary Table 1 (S1 Table).