Reach out to me for a PDF of the research paper
Healthcare Leadership Research | Interactive Evidence Dashboard
The Next Healthcare Capacity Crisis Is Diagnostic, Not Bed-Based
A decision-grade view of diagnostic access, completion, interpretation, clinical closure, equity, resilience, and the workflows that make physical capacity productive.
Staffed beds remain essential, but bed metrics observe only part of the production system. Diagnostic queues can constrain care before admission, during hospitalization, and after a result is issued.
01 | Executive overview
Capacity is the ability to turn clinical need into a usable decision
Bed management should not be replaced. It should be paired with governance of the diagnostic work that makes beds, clinics, and distributed networks clinically productive.
Dual-capacity leadership lens
| Conventional lens | Diagnostic lens |
|---|---|
| How many beds are staffed? | How many appropriate pathways reach closure by urgency? |
| What is occupancy? | Where is unresolved diagnostic work accumulating? |
| What is average length of stay? | Which diagnostic waits consume avoidable patient-hours? |
| How quickly are reports finalized? | How quickly do time-sensitive findings become action? |
| How many scanners and specialists exist? | How reliable is the end-to-end system by time, site, payer, neighborhood, and disease? |
Diagnostic capacity is an independent system constraint. The evidence does not establish that it is universally more limiting than staffed beds in every setting.
02 | Evidence atlas
Distinct studies converge on the same production constraint
Use the tabs to examine demand, interpretation, access, closure, inpatient flow, and distributed capacity. Denominators differ, so the values are not interchangeable benchmarks.
Projected modality growth
Population growth accounts for an estimated 73%-88% of projected increases, with aging accounting for 12%-27%.
Mean turnaround in 2024
The rise from approximately 2 h 11 m in 2014 is consistent with a system losing slack, particularly after 2021.
- Investigate work complexity, staffing, coverage, interruptions, and technology friction.
- Stratify by urgency, modality, subspecialty, time, and site.
- Pair speed with quality, workload, and report-to-action measures.
Turnaround is a partial signal. It does not measure accuracy, clinical action, or patient outcome.
The queue has a social pattern
Among 97,160 unique patients with MRI orders expected within one day, 48% were performed more than 10 days beyond the expected date.
Adjusted odds ratios show association, not causation. The study came from one quaternary system.
A signed report is not closure
Closure occurs when the responsible clinician reviews the result, communicates it appropriately, completes or documents the next action, and resolves ownership.
The three displayed rates come from different studies and populations. They illustrate failure modes and must not be treated as a common benchmark.
Beds and diagnostics are coupled constraints
About 30% of observed patient-days contained a delay. Internal hospital-system waits, including imaging and procedures, represented 33% of recorded delays.
The association does not prove that removing one wait hour will mechanically remove the same amount of length of stay.
Manage the pathway, not the scanner address
Clinicians estimated that roughly 10 percentage points of hospital-based commercial and Medicare care could shift to alternative sites using current technology.
- Reconcile external orders, results, and completion into one work-in-process view.
- Protect ownership, accreditation, communication, and closure across sites.
- Evaluate total episode performance, not local site cost alone.
The shift and savings values are scenario-based and may overstate realizable savings when fixed costs and network fragmentation persist.
03 | Effective capacity simulator
Usable capacity behaves like a weak-link production system
Adjust the five factors. The model demonstrates how apparently strong individual domains can compound into much lower end-to-end capacity.
Effective Capacity = P x W x R x C x L
When several factors are equal, improvement requires pathway-level testing rather than assuming that one asset is the answer.
04 | Diagnostic Capacity Index studio
Use the composite to reveal the profile, never to hide it
The provisional weights are held constant. Enter illustrative domain scores to see the weighted composite and the domains that need attention.
Priority: Equity and Completion
Interpret the number only with its six-domain profile, data-quality flags, confidence, and narrative explanation.
The DCI has not undergone psychometric or operational validation. Do not use it for public rankings, incentives, or interfacility league tables without formal validation.
05 | Workflow-first design lab
Replace the product-first reflex with a disciplined operating sequence
Workflow-first does not reject capital, AI, or outsourcing. It determines whether they address the binding constraint and verifies that they improve the complete pathway.
- Select a productStart with a scanner, platform, AI tool, or vendor offer
- Assume the use caseInfer value from feature claims or a local metric
- Deploy broadlyChange the operating model after implementation
- Discover the next queueAuthorization, staffing, interpretation, or closure becomes the new bottleneck
- Define the clinical outcomeChoose a sentinel pathway and urgency standard
- Map stages, timestamps, and ownersExpose all work from need recognition through closure
- Stratify the queueSegment by urgency, site, time, payer, geography, and equity
- Find the binding constraintDistinguish physical, workforce, workflow, completion, and closure failures
- Select a targeted interventionPair the intervention with quality, equity, and workforce guardrails
- Pilot and verifyMeasure completion, closure, downstream action, flow, and unintended effects
- Scale the proven responseUse capital or technology when evidence shows a residual constraint
Describe the operating problem
Close the loop on an abnormal cancer screening pathway
06 | Board governance and execution
Govern exceptions, owners, and recovery decisions
The board needs a small set of decision-grade signals. Operating detail should support the narrative without turning the board packet into a departmental scorecard.
| Board question | Primary signal | Required stratification | Trigger for action |
|---|---|---|---|
| Can patients enter the pathway on time? | Urgency-concordant order-to-completion | Disease, modality, site, payer, geography, deprivation | Sustained miss or widening high-risk tail |
| Is expert attention keeping pace? | Unread work aging and turnaround by urgency | Modality, subspecialty, time of day, site | Backlog growth plus quality or workload signal |
| Does information become action? | Time-sensitive abnormal-result closure | Setting, service, finding type, patient group | Unowned or overdue episodes |
| Are gains equitable? | Absolute and relative access and closure gaps | Race and ethnicity, language, payer, disability, geography | The gap widens despite average improvement |
| Can the system absorb disruption? | Recovery time, redundancy, vacancy, downtime readiness | Site and critical pathway | Recovery exceeds tolerance or a single point of failure |
| Is capacity producing value? | Avoidable patient-days, duplicate testing, leakage, margin, safety-net return | Service line and site | Cost rises without completion, quality, or access gain |
Implementation roadmap
Joint board oversight
Quality and strategy committees review diagnostic capacity at least quarterly.
One executive sponsor
Operational ownership stays distributed, but cross-functional accountability does not.
Enterprise capacity council
Include clinical, diagnostic, flow, digital, finance, equity, and patient representatives.
Pathway ownership
Service lines own targets and experiments; the enterprise standardizes definitions and data.
Decision-ready red signals
Every exception names an owner, cohort, safety control, recovery date, and verification measure.
Establish an enterprise diagnostic capacity council, baseline three sentinel pathways, and bring a one-page exception report to the board. Use the DCI as an internal design framework and validate it before formal adoption.
07 | Evidence, methods, and limitations
Use the dashboard for inquiry and governance, not false precision
The underlying report is a structured rapid evidence synthesis. It is decision-ready for internal leadership use but is not a systematic review, meta-analysis, causal model, or validated external benchmark.
For the existence and operational importance of diagnostic capacity constraints.
For selected access, closure, communication, and inpatient-flow failures.
Diagnostic and bed constraints are context-dependent and interdependent.
Weights, thresholds, case mix, reliability, responsiveness, and gaming risk require testing.
Studies most directly informing the hypothesis
| Study | Design / setting | Finding used | Principal limitation |
|---|---|---|---|
| Newman-Toker et al., 2024 | National modeling | Estimated 795,000 serious misdiagnosis-related harms annually | Not a capacity estimate; heterogeneous inputs |
| Christensen et al., 2025a | National insured claims plus Census projections | Imaging projected 16.9%-26.9% higher by modality in 2055 | Depends on utilization and insurance assumptions |
| Drake et al., 2026 | National Medicare outpatient claims | Mean turnaround 6.0 hours in 2024, 177% above 2014 | Timestamps do not measure accuracy or clinical action |
| Lacson et al., 2024 | 97,160 unique outpatient MRI orders | 48% delayed over 10 days beyond expected; patterned inequity | Single system and expected-date field |
| Ciemins et al., 2024 | 38 organizations; 20,581 adults | 47.9% completed colonoscopy within 6 months | External colonoscopies may be missed |
| Kapoor et al., 2023 | Closed-loop imaging follow-up program | 74.8% of clinically necessary recommendations completed | Single-system observational design |
| Bartsch et al., 2023 | 73,107 hospitalizations across five hospitals | Each imaging wait hour associated with 0.4-1.2 additional LOS hours | Observational, not causal |
| Nadler et al., 2026 | Day-level review of 1,152 inpatient days | 30% of days had delays; internal system delays were 33% | Single service and modest sample |
| Sahni et al., 2024 | 1,069-clinician survey linked to 2019 claims | About 10 percentage points of hospital volume potentially shiftable | Scenario-based; fixed costs may reduce savings |
| Abbasi et al., 2025 | Eight-year intervention-control study | Multifaceted program improved recommendation resolution | Additive interventions and one health system |
Interpretive safeguards
- Association is not causation, and not every delay causes harm.
- Use disease- and urgency-specific clocks rather than one universal speed target.
- Pair timeliness with appropriateness, diagnostic quality, patient choice, equity, and workforce sustainability.
- Display DCI domain scores, confidence, and data-quality flags alongside any composite.
- Validate definitions, thresholds, responsiveness, case mix, and susceptibility to gaming before external use.