Integrated Analytics Readout · Regional Health System

01 · The Executive Question

The decision this readout enables.

Every Integrated Analytics Readout exists to enable a specific call. The question is named before any synthesis begins — otherwise the readout drifts into a research summary. Below is the question this readout was commissioned to answer.

Executive Question Year 2 of 3 Transformation Health Check

“Is our value-based care program on track — or is reported adoption outrunning behavioural change and measured outcome improvement?”

Reader

CEO · CMO · CTO

Time Horizon

FY+1 reset · 6 months

Streams Required

All three · mandatory

Mind-Changing Evidence

Behavioural-quantitative gap

02 · The Three Streams

Three ways of knowing. One readout.

Every IAR pulls from the same three input families: qualitative interviews, AI-engine outputs, and quantitative benchmarks. The composition varies by engagement; the families do not. Below is the prepared input set for this readout.

01

◇ Stream 01 · Qualitative

Interview Data

Coded themes from 22 stakeholder conversations across clinical leadership, frontline physicians, care-management nurses, and ambulatory operations. Patient interviews not in scope for this readout cycle.

Interviews22

Coded themes14

Inter-rater reliability0.81 κ

AnonymityRole-aggregated

02

◇ Stream 02 · AI-Engine

AI-Engine Outputs

Pattern detection across cohort outcome trajectories, claim utilization signals, and care-management workflow telemetry — produced by the OneMindStrata intelligence engine. PHI-handled tier · no patient-traceable surfaces in the readout.

Models run6

Cohorts analyzed11

Confidence band0.78 — 0.93

PHI tierHandled

03

◇ Stream 03 · Quantitative

Benchmark Data

Internal performance against pre-program baseline plus a 9-system regional peer cohort. Quality, utilization, attribution-cohort outcomes, and total-cost-of-care — all normalized for risk and case-mix.

Metrics tracked18

Peer set9 systems

PeriodFY-2 to FY+0

NormalizationRisk + case-mix

❖

If two streams agree, the finding compounds. If two streams disagree, the disagreement itself is the finding. Single-stream signals are labelled — never presented with triangulated authority.

03 · The Synthesis Map

Where streams converge, where they disagree, and where only one carries the signal.

The synthesis pass produces three distinct buckets. Convergence is the strongest evidence in the readout; divergence is the most operationally important; single-stream findings are flagged so the reader can interrogate them.

Convergence

All three streams agree.

5

F-01 · Strongest signal

Care-management capacity is the binding constraint. Clinicians name it, the engine shows it in cohort touchpoint frequency, benchmarks confirm it versus peer systems.

I-03I-07 E-02E-04 B-01B-09

F-02 · Triangulated

Diabetes cohort outperforming. Care teams describe a coherent protocol, engine surfaces sustained A1C improvement, peers lag on the same cohort.

I-04 E-01 B-03

F-03 · Triangulated

Specialty referral leakage is rising. Operators see it, the engine flags increasing out-of-network claims, benchmarks place leakage rate above peer median.

I-09 E-05 B-12

F-04 · Triangulated

Documentation burden is up, not down. Physicians describe it, engine telemetry shows after-hours EHR time rising, peer systems with similar tooling are also rising.

I-11 E-03 B-15

F-05 · Triangulated

Total cost of care is moving in the right direction — modestly. Leadership describes the trend, engine confirms PMPM trajectory, benchmarks place the system in middle quartile against peers.

I-01 E-06 B-04

Divergence

Streams disagree.

3

F-06 · Reported > observed

Risk-stratification adoption is reported high; behaviour says otherwise. Clinicians say the tool is fully embedded; engine shows only 47% of high-risk patients receive the recommended touchpoint within 14 days.

I-05I-12 E-07

F-07 · Performance contested

Quality metrics show improvement; clinicians describe degradation. Benchmarks place HEDIS scores above peer median; clinician interviews describe workflow strain reducing time per patient. Trust the benchmark for the contract; trust the interviews for the ops plan.

I-08I-14 B-05B-08

F-08 · Source disagreement

Patient panel size: depends on which system you ask. Engine pulls from EHR attribution; benchmarks pull from payer attribution; the gap is 8% systematically. Reader should reconcile before any panel-based action.

E-08 B-11

Single-Stream

Only one source carries it.

4

F-09 · Interview only

Care managers report a confidence problem with the new escalation pathway. Theme is consistent across 11 of 22 interviews; engine and benchmarks are silent. Treat as priority signal — benchmarks lag culture by quarters.

I-06I-13

F-10 · Engine only

Statistically significant variance in admission patterns by ZIP code. Engine surfaced it; clinicians did not raise it; benchmarks are not granular to that level. Worth investigation but not yet actionable.

E-09

F-11 · Benchmark only

Length-of-stay variance versus peers is widening for one DRG cluster. No interview signal; engine did not flag. Likely a coding artifact; verify before acting.

B-14

F-12 · Interview only

Physicians describe a perceived widening gap between primary care and specialist capacity. Anecdotal in interviews; not yet visible in claim data or peer benchmarks. Watch in next cycle.

I-15

04 · Drivers

What is actually moving performance — in either direction.

Drivers explain current performance. Each driver is sized in the unit the reader cares about: contract dollars, patients, basis points of margin, weeks of cycle time. Confidence is reasoned, not asserted.

01

◆ Driver · Positive Confidence · Strong

Care-management cohort discipline is converting risk-adjusted spend into shared-savings yield.

Three streams confirm it. The diabetes-cohort protocol is showing a sustained A1C improvement curve against the pre-program baseline; engine telemetry shows structured outreach actually happening, not just documented; peers running similar programs lag on the same cohort. This is the program's strongest evidence of working as designed.

Provenance I-04 E-01 B-03

Magnitude

+18% A1C control
vs baseline

02

◆ Driver · Positive Confidence · Strong

Total cost of care trajectory is improving — modestly — against pre-program baseline.

PMPM trend has bent. Year-over-year, the attributed-population PMPM is down 2.1% against an inflation-pressured backdrop in which peer systems averaged a 1.6% increase. The improvement is real and durable; it is also smaller than the program plan assumed.

Provenance I-01 E-06 B-04

Magnitude

−2.1% PMPM
vs +1.6% peers

03

◆ Driver · Negative Confidence · Strong

Specialty referral leakage is actively eroding shared-savings yield.

Operators see it; engine flags increasing out-of-network claims; benchmarks place the system above peer-median leakage rate. Each percentage point of leakage in the attributed cohort corresponds to roughly $2.4M of annualized contract value moving outside the network. Three streams agree on direction and magnitude.

Provenance I-09 E-05 B-12

Magnitude

~$2.4M / pp
annualized

04

◆ Driver · Negative Confidence · Medium

Documentation burden is up, not down — and it is converting clinician hours into administrative hours.

Physicians describe it as the dominant frustration of Year 2. Engine telemetry confirms after-hours EHR time rising 11% year-over-year. Peer systems with similar tooling stacks are seeing the same direction, suggesting this is partly an industry pattern, partly system-specific. Confidence medium because attribution between program-driven and tool-driven causes is unresolved.

Provenance I-11 E-03 B-15

Magnitude

+11% YoY
after-hours EHR

05 · Gaps

Closeable distance between current state and target state.

Gaps are not problems — they are sized distances. Each gap names what is missing, how big the gap is in the reader's units, and what would close it. Gaps without sizing are not actionable.

01

× Gap · Operational Confidence · Strong

Care-management capacity is the binding constraint on cohort expansion.

All three streams agree. The current panel-to-care-manager ratio is ~340:1; benchmark median for systems hitting their VBC targets is ~245:1. Closing the gap requires either expanding capacity by ~38% or shrinking the actively-managed cohort. The program's growth plan assumes the former; current hiring trajectory has not started.

Provenance I-03 I-07 E-02 B-01

Gap Size

340:1 vs 245:1
+38% capacity

02

× Gap · Behavioural Confidence · Strong

Risk stratification: reported adoption is outrunning actual workflow change.

Clinician self-report places adoption near 90%; engine shows only 47% of high-risk patients receive the recommended 14-day touchpoint. The gap is roughly 43 percentage points. This is the most important divergence in the readout because it answers the executive question directly — reported adoption is materially outrunning behavioural change.

Provenance I-05 I-12 E-07

Gap Size

~90% reported
vs 47% observed

03

× Gap · Strategic Confidence · Medium

Specialty network completeness is insufficient for the attributed cohort's needs.

Driver 03 surfaces leakage as a symptom; the underlying gap is network design. Cardiology and behavioural-health capacity in-network sit below cohort utilization patterns by an estimated 15–22%. Closing the gap is partly contracting (referral relationships) and partly capacity (employed specialists). Confidence medium because the gap-sizing depends on attribution method.

Provenance E-05 B-12 B-16

Gap Size

15–22% short
specialty capacity

04

× Gap · Cultural Confidence · Medium

Care managers do not yet trust the new escalation pathway.

CMIO · Clinical Informatics

Horizon

Begin FY+1 Q3 · 6 months

Success Signal

After-hours EHR time flat YoY by FY+2

Priority

C · Third

07 · Confidence Ledger

What the reader should act on, what to watch, and what to verify.

Every finding in the readout carries a confidence band. Strong findings are decision-grade; medium findings inform the plan but should be tested; weak findings are watch-list, not action-list.

◆ Strong

7

Three streams agree, with independent paths. Decision-grade. Act.

F-01 · Care-management capacity is binding

F-02 · Diabetes cohort outperforming

F-03 · Specialty referral leakage rising

F-05 · Total cost trajectory improving

D-01 · Cohort discipline yielding savings

D-02 · PMPM trend bent versus peers

D-03 · Leakage eroding savings

◆ Medium

5

Two streams agree, or attribution is partly unresolved. Inform plan; test before scaling.

F-04 · Documentation burden up

F-06 · Reported > observed adoption

F-07 · Quality vs clinician sentiment split

G-03 · Specialty network insufficient

G-04 · Care-manager pathway distrust

◆ Weak

4

Single-stream signal. Watch — do not yet act. Re-test in next cycle.

F-08 · Panel attribution gap

F-09 · Care-manager confidence (interview)

F-10 · ZIP-level admission variance (engine)

F-11 · LOS variance one DRG (benchmark)

08 · Provenance Manifest

Every finding traces back to numbered manifest items.

The manifest is the auditability layer. Without it, a synthesis is opinion. Below is the abbreviated manifest underlying this readout. Full manifest archived alongside the readout per Stratenity Internal SOP.

◆ Manifest excerpt · v1.0

22 interviews · 6 engine outputs · 18 benchmark metrics — numbered, dated, source-attributed.

◇ Stream 01 · Interviews

I-01

Exec leadership · PMPM trajectory framing

I-03

VP Care Mgmt · capacity as binding

I-04

Endocrinology · diabetes protocol

I-05

PCP cohort · risk-strat self-report

I-06

Care managers · pathway trust

I-07

Care managers · ratio strain

I-09

Ambulatory ops · referral patterns

I-11

PCP cohort · documentation strain

I-12

PCP cohort · adoption claim

I-13

Care managers · escalation friction

◇ Stream 02 · OneMindStrata

E-01

Cohort Outcome v3.2 · A1C trajectory

E-02

Workflow Telemetry v1.8 · touchpoint frequency

E-03

EHR Time-Use v2.1 · after-hours rising

E-04

Cohort Outcome v3.2 · touchpoint vs target

E-05

Claim Pattern v2.0 · out-of-network trend

E-06

PMPM Trajectory v1.4 · attributed cohort

E-07

Workflow Telemetry v1.8 · 14-day completion

E-08

Attribution v1.0 · EHR-derived panel

E-09

Geo Variance v0.9 · ZIP-level admission

◇ Stream 03 · Benchmarks

B-01

Peer cohort · panel:CM ratio

B-03

Peer cohort · diabetes outcome quartile

B-04

Peer cohort · PMPM YoY

B-05

HEDIS scorecard · quality measures

B-08

HEDIS scorecard · quality trajectory

B-09

Peer cohort · care-mgmt capacity median

B-11

Payer attribution · panel definition

B-12

Peer cohort · leakage rate median

B-14

Peer cohort · LOS by DRG cluster

B-15

Peer cohort · EHR after-hours time

B-16

Peer cohort · specialty capacity in-network

Persistence Panel

What becomes durable beyond this readout.

Every IAR produces artifacts that persist beyond the engagement and feed the next one. The Stratenity standard requires the persistence panel as the closing element — knowledge stays and compounds, even when consultants leave.

01

Provenance manifest

Every finding traceable back to numbered items. Underwrites the next IAR cycle and anchors auditability for board and regulator.

02

Synthesis card library

12 synthesis cards covering convergence, divergence, and single-stream findings — reusable as the starting set for the FY+1 quarterly check-in.

03

Confidence ledger

Strong / medium / weak band on every finding, with reasoning. Carries forward to the next cycle — weak findings re-tested, medium findings re-classified as evidence accumulates.

04

Action watch list

5 sequenced actions with owners, horizons, and success signals. The next IAR cycle measures movement against these signals — not against unspecified intent.

Where reported adoptionmeets actual change.

The decision this readout enables.

Three ways of knowing. One readout.

Interview Data

AI-Engine Outputs

Benchmark Data

Where streams converge, where they disagree, and where only one carries the signal.

What is actually moving performance — in either direction.

Care-management cohort discipline is converting risk-adjusted spend into shared-savings yield.

Total cost of care trajectory is improving — modestly — against pre-program baseline.

Specialty referral leakage is actively eroding shared-savings yield.

Documentation burden is up, not down — and it is converting clinician hours into administrative hours.

Closeable distance between current state and target state.

Care-management capacity is the binding constraint on cohort expansion.

Risk stratification: reported adoption is outrunning actual workflow change.

Specialty network completeness is insufficient for the attributed cohort's needs.

Care managers do not yet trust the new escalation pathway.

Sequenced moves. Owners. Horizons. Signals of success.

Reset the care-management capacity plan and the escalation pathway protocol together — before any cohort expansion.

Close the reported-versus-observed adoption gap by changing what gets measured.

Stand up a specialty-network closure plan with a clear build-versus-contract sequence.

Reconcile the panel-attribution gap before any panel-based action goes operational.

Convene a documentation-burden working group — but only once Sequence A has cleared.

What the reader should act on, what to watch, and what to verify.

Every finding traces back to numbered manifest items.

22 interviews · 6 engine outputs · 18 benchmark metrics — numbered, dated, source-attributed.

What becomes durable beyond this readout.

Provenance manifest

Synthesis card library

Confidence ledger

Action watch list

Where reported adoption
meets actual change.