Health system reform runs into the same barriers across countries and care models
Health systems are being asked to do more than ever before: manage non-communicable diseases, care for older populations, and respond to chronic workforce shortages while still delivering timely, coordinated treatment. The central tension is structural. Most health systems were built around episodic acute care, yet the greatest demand now comes from patients who need prevention, early detection, long-term follow-up, and services that are easier to reach than a tertiary hospital.
A mixed-methods review published in Frontiers in Health Services examined how countries are trying to reset that balance. The focus fell on three domains that now dominate reform agendas: service delivery decentralisation, workforce transformation, and screening and diagnostic services for early detection and treatment. Across those domains, the work identified common barriers and enablers that recur regardless of geography, financing model, or health system size.
The message from the evidence is blunt.
Transformation fails less often because the idea is wrong than because politics, incentives, workforce culture, and operational design pull in different directions.
67 transformation cases, 34 barriers, and 27 enablers across six system domains
The review combined a scoping literature search with semi-structured one-on-one interviews. Searches were conducted in EMBASE, Global Health, and Ovid MEDLINE, with reference management, deduplication, and screening handled through the Rayyan platform. Two members of the research team independently reviewed titles and abstracts before moving to full-text review and data extraction. In total, 153 peer-reviewed papers and 35 grey literature reports were analyzed, and 17 expert stakeholders from six countries were interviewed, including policy experts, academics, health policy experts, and clinicians.
From that evidence base, 67 transformation case studies were included. More than half focused on service delivery decentralisation, with 36 cases in that category. Screening accounted for 16 cases, while 12 addressed workforce adaptation. The distribution itself is revealing: health system reform is not happening as a single integrated redesign. It is happening in separate streams that often depend on one another but are managed through different agencies, budgets, and professional hierarchies.
The review then distilled 34 barriers and 27 enablers, grouping them into six areas: political, governance and policy; funding, incentives, and resource allocation; workforce capacity, capability, and culture; system integration, structure and care coordination; digital infrastructure, technology and data; and communication, engagement, and change management.
Political commitment matters, but policy design and execution determine whether reform survives contact with reality
Political support can open the door to reform, yet durable change depends on whether policy is translated into usable operating rules. When responsibility shifts from central institutions to local care settings, the system has to answer practical questions immediately: who pays, who staffs, who measures quality, and who is accountable when referrals stall or diagnostics are delayed. Without those answers, decentralisation can produce fragmentation instead of access.
Funding and incentives emerged as a separate pressure point because many transformation programs ask providers to work differently while payment structures still reward old behavior. A hospital-based system may remain financially dependent on activity volumes even as policymakers push care into the community. In that setting, even well-intentioned reforms can meet silent resistance, not because clinicians oppose better care, but because the new workflow threatens existing budgets, staffing patterns, or professional status.
The review also captured a recurring governance issue: complex reforms need coordination across institutions that do not naturally move in sync. Screening programs, diagnostic pathways, primary care, specialist services, and workforce planning can each be improved in isolation, yet patient access depends on the handoff points between them. When those handoffs are weak, a reform may look successful on paper while remaining difficult to experience at the bedside.
Workforce capacity is not only a staffing problem; it is a culture and design problem
Workforce transformation ran through the evidence as more than a numerical shortage. Capacity includes training, role clarity, confidence with new tasks, willingness to adopt different team structures, and the presence of local leadership that can normalize change. A new care model can fail when staff are asked to expand their scope without the support, supervision, or operational tools needed to do the job safely.
Culture also shapes whether change is absorbed or resisted. Teams accustomed to hierarchical referral pathways may struggle with decentralized care if authority, responsibility, and clinical ownership become blurred. In practice, that can slow implementation even when the underlying policy is sound. The strongest programs are likely those that align new roles with real clinical demand and avoid placing additional burdens on already stretched staff without removing outdated tasks elsewhere.
Digital infrastructure appears in the review as another enabling layer, but technology alone cannot repair workflow. Data systems matter because they allow continuity across settings, support triage, and make screening and follow-up visible. Yet if data platforms do not integrate cleanly across organizations, the result can be more administrative burden rather than better care coordination.
Screening and decentralization succeed when patients can actually move through the system
Early detection depends on more than a screening test. It also depends on whether patients can get to the test, receive the result, and enter treatment without being lost between services. Decentralized delivery can shorten the path to care, but only if referral networks, laboratory systems, and clinical oversight are built to support it. Otherwise, the point of contact moves closer to the patient while the bottleneck simply shifts downstream.
The review’s six domains reflect how tightly linked these issues are. Communication and engagement shape trust and uptake. Governance determines whether reforms persist beyond pilot phase enthusiasm. Resource allocation determines whether changes are sustainable. Care coordination determines whether a patient can move from suspicion to diagnosis to treatment without avoidable delay. Each layer influences the next.
A second standalone reality sits underneath the numbers.
A screening program that cannot deliver treatment is not early detection; it is deferred frustration.
Twenty-five recommendations aim to make reform more scalable and context-sensitive
The final output of the review was a set of 25 recommendations organized into seven groups for policymakers and other stakeholders. The recommendations were framed around making transformation more scalable and more responsive to local context. In practical terms, that means reform must be planned as a system exercise rather than a single-policy announcement. Leadership, financing, workforce development, digital tools, and community engagement need to be designed together if change is expected to last.
For health systems under pressure from ageing populations and chronic disease, the value of this review lies in its cross-cutting map of failure points and success conditions. It does not offer a single universal reform template, and the evidence base spans distinct countries and program types. It does, however, show where reform efforts repeatedly stall and what tends to support progress when systems attempt to move care closer to home, expand screening, or redesign workforce roles.
The evidence leaves one operational question in sharp relief: whether governments and health leaders can align financing, governance, and workforce support quickly enough to match the scale of need. The review provides the checklist, but implementation will decide which systems can use it.