Research and Model Design by Kelly Emrick, DHSc, PhD, MBA, BSRT(ARRT)R
The Emrick Structure–Process–Outcomes (SPO) Model
From slogan to working model
SPO begins with a memorable idea — Structure + Process + Outcomes = Success — but the rigorous reading is not literal arithmetic. Structure creates capability and constraint; process converts capability into performed work; outcomes reveal what that work produced; and success is a governed value function, not a vague label.
Capability before the encounter
Durable organizational attributes that shape capacity and constraint: staffing, equipment, facilities, information systems, protocols, governance, capital, and safety infrastructure.
Capability into reliable work
The care actions and coordination actually performed: handoffs, diagnostic flow, standard work, response times, protocol adherence, and recovery from variation.
The consequences of care
Clinical, operational, financial, patient-reported, workforce, and equity-sensitive effects — mortality, infections, readmissions, LOS, HCAHPS, margin, turnover, subgroup gaps.
Core contributions
The causal-operational architecture
Structure shapes process capability; process reliability mediates outcomes; outcome data feeds the learning loop that re-shapes structure.
The three linked domains
SPO inherits Donabedian’s structure–process–outcome logic and Lean, Six Sigma, and systems-thinking traditions, organized into a common leadership grammar without reducing healthcare to an industrial metaphor.
Durable capability
Relatively durable attributes that shape capacity and constraint before the episode of care begins.
Leadership indicators
- Staffed beds, exam rooms, equipment capacity
- RN hours per patient day, physician coverage
- Digital maturity, protocol ownership
- Safety infrastructure, governance cadence
Units counts · ratios · FTEs · coverage hours · maturity scores
Reliable performed work
The care actions, workflows, and reliability patterns actually performed during service delivery.
Leadership indicators
- Door-to-provider, order-to-result
- Bundle adherence, handoff completeness
- Discharge readiness, scheduling reliability
- Authorization cycle time, rework rate
Units minutes · percent adherence · rates/episode · defects/opportunity
Consequences of care
Clinical, operational, financial, patient-reported, workforce, and equity-sensitive effects.
Leadership indicators
- Mortality, complications, CLABSI/CAUTI/SSI
- Readmissions, LOS, left-without-being-seen
- HCAHPS, PROMIS, margin, turnover
- Equity gaps across subgroups
Units risk-adjusted rates · days · dollars · survey scores · SIRs
The minimum working model
A framework becomes a model when it defines variables, specifies pathways, and states what counts as success — including explicit tradeoffs. The structural form preserves the slogan while preventing double-counting.
Pathway equations
Process is a function of structure and context; outcomes are a function of structure and process and context; success is a weighted value function over the outcome vector. The weights are governance choices, not statistical artifacts.
Composite scoring logic
Metrics are normalized 0–100 with directionality fixed in advance. The composite is a navigational instrument for executive communication — it never replaces the pathway model or clinical judgment.
SPO Index calculator
Normalize each domain metric to 0–100 (directionality already applied: higher is always better here). Domain scores combine with explicit weights and a balancing-measure penalty. Adjust the θ weights to reflect your organization’s emphasis.
Domain weights & balancing penalty
θ weights are auto-normalized. The penalty deducts points when balancing measures (burnout, equity gap, downstream utilization) deteriorate.
Domain profile
Weighted contribution to the index
Define success before you measure it
Success is a weighted combination of value domains minus penalties. The weights are a governance decision: if success means lower wait time, one intervention wins; if it means short-run margin, another wins; if it includes equity and workforce sustainability, the best decision may change again. This tab makes that choice explicit.
Value-domain scores (0–100)
Governance weights
What each domain contributes to Success
Queueing Lab: structure vs. process levers
In an M/M/s queue, structure maps to the number of servers (s) — clinicians, rooms, scanners, beds — while process maps to the effective service rate (μ). Standard work, reduced rework, and faster diagnostics raise μ without adding capacity. The lab lets you compare both levers in one operational frame.
Inputs
Baseline configuration. Move the sliders to test a structure or process lever.
Live results
The four SPO scenarios
Recomputed live for your chosen horizon. With the default 12-hour, λ = 8.5 settings these reproduce the manuscript’s illustrative table.
Access effect — expected wait
Adding a server sharply reduces waiting; the combined lever yields the shortest queue.
Fiscal trade-off — modeled net value
In this illustrative value function, process improvement creates the highest net value by avoiding recurring labor cost.
| Scenario | Structure | Process μ | Wait (min) | LWBS | Completed | Net value |
|---|
The mature SPO analytics stack
The model becomes useful when measurement, causal estimation, operations modeling, and governance are connected. Match the method to the question — and respect each method’s caution.
Multilevel regression
Risk-adjusted benchmarking across units, clinicians, service lines, or time.
Caution: association is not causal leverage when confounding remains.
Causal DAGs & mediation
Separate direct structural effects from process-mediated effects.
Caution: bad adjustment creates bias; the causal question must be explicit.
Structural equation modeling
Model latent constructs — safety climate, coordination, digital maturity.
Caution: model fit is not proof of causal truth.
Queueing & network flow
Test access, waits, bottlenecks, staffing patterns, and utilization.
Caution: real workflows violate simplified assumptions.
State-space & control
Manage dynamic demand–capacity: occupancy, boarding, surge.
Caution: harder to explain, but valuable for command centers.
Simulation & optimization
Test expensive or irreversible interventions before implementation.
Caution: without validation, simulation becomes a polished story.
Enterprise measurement architecture
Most organizations already hold the data — it just lives in disconnected systems. SPO turns these fragments into a reproducible data product with common identifiers, denominator rules, source lineage, and version-controlled logic.
| Data layer | Representative variables | Source systems | Use in the model |
|---|---|---|---|
| Clinical events | Diagnoses, vitals, meds, labs, imaging, procedures, discharge status | EHR, lab, imaging, pharmacy | Risk adjustment, outcome definition, process timestamps |
| Flow & access | Arrival, triage, roomed, seen, order, result, admit, discharge, transfer | ADT, ED systems, bed mgmt, scheduling | Queueing models, throughput, congestion, delay |
| Safety surveillance | Device days, infection events, procedure classes, predicted counts | NHSN, infection-prevention software | Risk-adjusted safety outcomes & harm surveillance |
| Experience & PROMs | HCAHPS, CAHPS, PROMIS, complaints, access barriers | Survey platforms, portals, call centers | Patient-reported outcomes, experience, equity, access |
| Workforce | Hours, overtime, agency use, turnover, sick calls, skill mix, vacancy | HRIS, payroll, scheduling | Structural capability & balancing measures |
| Finance | Cost/case, contribution margin, denial rate, payer mix, labor & supply cost | Cost accounting, billing, GL, contracts | Fiscal performance & intervention return |
Leading
Structure & process measures — change before outcomes become visible.
Lagging
Outcomes — require enough cases and time to detect stable change.
Balancing
Protect against local optimization — burnout, equity gaps, downstream utilization.
Implementation roadmap
SPO is implemented as a governed learning system, not a one-time dashboard build. Define the enterprise question in value terms, build a governed family of measures, create a reproducible data product, run in shadow mode, pilot with counterfactuals, then scale.
Charter
Define the service line, primary outcome, process measures, balancing measures, and the success function.
Data audit
Validate timestamps, denominator logic, data lineage, missingness, and case-mix variables.
Baseline modeling
Estimate baseline pathways with hierarchical, causal, queueing, or simulation methods as appropriate.
Shadow mode
Run the model during testing without operational consequences — assess calibration, drift, subgroup stability, and face validity.
Pilot
Implement the intervention with a pre-specified evaluation design and balancing measures.
Scale decision
Expand, modify, or stop based on outcomes, costs, safety, equity, and workforce effects.
Twelve-month cadence
An illustrative pilot → validation → scale decision timeline.
Governance safeguards for responsible use
A performance model is never neutral. The variables selected, weights assigned, groups stratified, and outcomes emphasized all reflect values. Governance must be explicit about what the organization refuses to sacrifice.
Metric tunnel vision
Failure mode: improving one visible metric while harming unmeasured outcomes.
Safeguard: use outcome, process, and balancing measures together.
Equity masking
Failure mode: average improvement hides deterioration for a subgroup.
Safeguard: stratify key outcomes; review equity deltas before scale.
Gaming
Failure mode: teams optimize documentation or coding rather than performance.
Safeguard: audit definitions, triangulate sources, add qualitative review.
False precision
Failure mode: outputs treated as certain despite uncertainty.
Safeguard: show confidence intervals, sensitivity analyses, limitations.
Workforce harm
Failure mode: throughput improves while burnout, turnover, or safety risk rises.
Safeguard: include workforce sustainability as a core balancing domain.
Privacy erosion
Failure mode: linked data used for punitive surveillance.
Safeguard: define access, governance, and learning-purpose boundaries in advance.
Limitations to hold in view
Structure, process, and outcomes are linked probabilistically, not mechanically. Outcomes can be delayed, confounded, and shaped by factors outside the organization.
Apparent improvement may reflect coding changes, regression to the mean, secular trends, or selection effects. Never deploy as a simplistic before-and-after scorecard without validation and counterfactual thinking.
SPO readiness assessment
Rate your organization across the four readiness domains. The radar reveals where capability is uneven and the tier suggests where to start — charter and data first, foundation building, or ready to pilot.
Rate each statement
Absent → Embedded
The Emrick Structure–Process–Outcomes (SPO) Model · interactive operating model derived from the manuscript by Kelly Emrick, DHSc, PhD, MBA, BSRT(ARRT)R.
Builds on Donabedian (1966/2005; 1988), improvement science, queueing theory (Green, 2006), causal inference (Pearl, 1995; Hernán & Robins, 2020; VanderWeele, 2015), and evidence streams including Pronovost et al. (2006), Haynes et al. (2009), and Aiken et al. (2002). Queueing and value figures are illustrative and configurable.
