⚕️ Microsoft Fabric Semantic Model Modes : A use case in the NHS-NHS ICBs vs NHS Trusts.

Choosing the right Power BI storage mode in Fabric is not a one-size-fits-all decision. NHS organizations—whether at the system level (ICBs) or the provider level (Trusts)—face distinct pressures around real-time operations, population health management, governance, and reporting cycles.

Here’s how Direct Lake, DirectQuery, Import, and Composite align with NHS use cases.

1. Import Mode – Batch Analytics & National Submissions

Source: Microsoft learn

Best Fit: ✅ NHS ICBs

Why:

• ICBs typically need to consolidate population-level datasets (Primary Care, Secondary Care, Community, Social Care).

• Data doesn’t change second-by-second; it’s more about monthly submissions, trend analysis, and planning.

• Import mode allows high performance dashboards on datasets like SUS, SLAM, GP data, and population health indicators.

ICB Use Case Examples:

• Population health management (PHM) risk stratification.

• Month-end reporting for NHSE submissions.

• Financial performance dashboards (planned care backlog, elective RTT trends).

Limitations for ICBs:

• Not suitable where real-time care coordination is required.

• Refresh cycles (daily/weekly) can create data lags.

  
  

2. DirectQuery Mode – Real-Time Operational Insights

Source: Microsoft learn

Best Fit: ✅ NHS Trusts

Why:

• Trusts manage frontline operational pressures where real-time visibility is critical (e.g., bed state, ED flow, sepsis detection).

• DirectQuery ensures live data from PAS/EPR/LIMS systems without waiting for refreshes.

• Maintains source-level security, useful in Trusts where data governance is strict at the patient level.

Trust Use Case Examples:

• ED & Acute bed occupancy dashboards.

• Theatre scheduling and utilisation.

• Real-time infection/sepsis monitoring from clinical systems.

Limitations for Trusts:

• Performance depends on the underlying system (many EPRs are not optimised for BI queries).

• Risk of overloading live transactional systems if poorly designed.

  
  

3. Direct Lake Mode – System-Wide, Scalable, Low-Latency

Source: Microsoft Learn

Best Fit: ✅ Both ICBs & Trusts (future-facing)

Why:

• Direct Lake in Fabric unlocks a shared, governed data layer (OneLake) across ICBs and Trusts.

• Combines Import speed with DirectQuery freshness—critical for ICB-Trust collaboration.

• Trusts benefit from near real-time data without burdening EPR systems.

• ICBs benefit from unified longitudinal datasets that are always fresh for PHM.

ICB Use Case Examples:

• Federated data platform integration (e.g., population cohorts with refreshed daily hospital feeds).

• Demand & capacity planning across multiple Trusts.

Trust Use Case Examples:

• Bed management, sepsis detection, and clinical dashboards with near real-time refresh.

• Integration with wearables or remote monitoring (IoT data streams).

Limitations:

• Requires Fabric adoption and migration of data pipelines to OneLake.

• Still maturing—some advanced modelling features may lag Import.

  
  

4. Composite Mode – Balancing System and Provider Needs

Source: Microsoft Learn

Best Fit: ✅ Hybrid scenarios (ICB + Trust collaboration)

Why:

• Allows blending historical data (Import) with real-time operational feeds (DirectQuery/Direct Lake).

• Critical where NHS needs both long-term planning and immediate operational oversight.

• Supports hybrid tables—e.g., import last 5 years of hospital activity, query last 24 hours in real time.

ICB Use Case Examples:

• Elective recovery dashboards: historical RTT trends + live waiting list feeds.

• Workforce planning: import HR data, query agency staffing in real time.

Trust Use Case Examples:

• ED dashboards: imported trends (last year volumes) + real-time occupancy (Direct Lake).

• Infection control: historic outbreak data + live patient infection flags.

Limitations:

• Complexity—requires strong governance and modelling discipline.

• Can confuse end-users if not communicated clearly (different refresh cadences).

🏥 NHS Alignment Summary

Mode	Best For	ICB Alignment	Trust Alignment
Import	Batch analytics, trend reporting	✅ Strong fit (population, planning, PHM)	⚠️ Limited (not real-time)
DirectQuery	Live operational data	⚠️ Less common	✅ Strong fit (bed mgmt, sepsis, ED flow)
Direct Lake	Fabric-native near real-time	✅ Strong fit (federated PHM, demand planning)	✅ Strong fit (clinical ops, capacity mgmt)
Composite	Mixed workload (historical + real-time)	✅ For backlog & workforce planning	✅ For ED/clinical ops with history + live feeds

✅ Key Takeaway:

• NHS ICBs: Best served by Import (planning, submissions, PHM) and Direct Lake (system-wide federated analytics).

• NHS Trusts: Best served by DirectQuery (real-time ops) and Direct Lake (next-gen real-time at scale).

• Composite: Acts as the bridge between ICB and Trust priorities, especially for integrated care planning.

Watch related video on my you tube channel below:

https://youtu.be/M7uMVULJA_Y