Aurelian’s Casebook
Casebook #2
A performance indicator can tell us something about a system. That does not make improving the indicator the purpose of the system.
Healthcare organizations measure utilization for good reasons.
Admissions, bed-days, occupancy, outpatient activity, and referrals help leaders understand demand, capacity, access, and how healthcare resources are being used.
But an indicator can quietly become something else.
Instead of helping leaders understand the system, improving the indicator itself becomes the objective.
That distinction is at the heart of this case.
The numbers are moving in the wrong direction.
Or so it seems.
Case Presentation
A regional healthcare system is reviewing three years of inpatient activity.
The executive dashboard shows a consistent pattern:
| Inpatient Indicator | Year 1 | Year 2 | Year 3 |
| Admissions | 8,400 | 7,700 | 7,050 |
| Inpatient bed-days | 39,500 | 35,900 | 32,600 |
| Average occupancy | 72% | 66% | 60% |
Leadership describes the pattern as declining inpatient productivity.
The concern is understandable. But the terminology already contains an assumption.
Admissions and bed-days describe inpatient activity. Occupancy describes the proportion of available bed capacity being used. None of these measures, by itself, establishes whether the hospital is becoming more or less productive in the broader sense.
Nevertheless, the management question appears straightforward:
How do we increase inpatient productivity?
Aurelian begins somewhere else.
Does this population actually need more inpatient care?
The investigation changes direction.
The Case File
Decision Under Review
Determine whether declining inpatient utilization represents institutional underperformance requiring corrective action.
Initial Evidence
- Declining admissions
- Declining inpatient bed-days
- Declining occupancy
Missing Evidence
- Population and demographic change
- Clinical demand
- Referral patterns
- Available specialty capabilities
- Outpatient management
- Access to local services
- Workforce and operational capacity
Central Analytical Question
Why is inpatient utilization declining, and does that decline represent improvement or deterioration?
Investigation Status
Multiple Hypotheses Under Review
Evidence Review
Exhibit A — Inpatient Utilization Is Declining
The initial dashboard is not wrong.
| Indicator | Three-Year Change |
| Admissions | −16.1% |
| Inpatient bed-days | −17.5% |
| Average occupancy | 72% → 60% |
Something has changed.
What the dashboard cannot tell us is why.
Declining utilization could reflect changing population need, different patterns of care, appropriate referral, excess capacity, or other structural changes.
It could also reflect something less favorable: reduced access, workforce shortages, operational barriers, inappropriate referrals, or patients seeking care elsewhere.
The numbers establish the finding.
They do not establish the explanation.
They are evidence, not diagnoses.
Exhibit B — The Population Has Changed
The demographic review provides the next clue.
The hospital’s service population has remained relatively stable overall, but its composition has changed.
Births have declined, younger age groups have contracted, and the proportion of older residents has increased.
Those changes do not automatically predict lower hospital use. Population ageing, for example, can increase demand for hospital care even as other demographic shifts reduce demand for particular services.
The important finding is not that demographic change explains the decline.
It is that the population generating healthcare demand is no longer identical to the population around which earlier patterns of hospital activity developed.
Historical utilization cannot automatically be treated as current demand.
Exhibit C — Referral Patterns Have Changed
The review next turns to patients transferred to higher-complexity centers.
| Referral Indicator | Year 1 | Year 2 | Year 3 |
| Transfers to tertiary centers | 410 | 485 | 560 |
| Transfers involving services unavailable locally | 62% | 71% | 79% |
The evidence does not prove that referrals account for the decline in admissions.
It does establish that referral behavior has changed.
More patients are reaching higher-complexity institutions, and an increasing proportion of those transfers involve services the local hospital does not provide.
Regionalized healthcare systems may deliberately concentrate specialized services and expertise in selected institutions. From the local hospital’s perspective, however, those patients represent inpatient activity occurring elsewhere.
Whether this represents good or poor performance depends on another question:
Were patients transferred because higher-level care was appropriate—or because local capability had deteriorated?
The utilization statistics cannot answer that question alone.
Exhibit D — Care Delivery Has Also Changed
The outpatient data reveal another shift.
| Care Setting | Year 1 | Year 2 | Year 3 |
| Inpatient admissions | 8,400 | 7,700 | 7,050 |
| Ambulatory / same-day encounters* | 14,200 | 15,600 | 17,100 |
Fictional case data representing services managed without overnight inpatient admission.
Inpatient activity has declined while ambulatory activity has increased.
Again, the finding does not establish causation.
But it makes one conclusion increasingly difficult to defend: inpatient utilization cannot be interpreted without considering where care is now being delivered.
Across health systems, some procedures and episodes of care have shifted toward day or same-day treatment rather than overnight hospitalization when clinically appropriate.
That may be part of what is happening here.
It remains a hypothesis, not a conclusion.
Exhibit E — A Competing Explanation
The investigation uncovers another trend.
Vacant clinical positions have increased during the same period, particularly in several services responsible for inpatient care. Some elective procedures are also experiencing longer scheduling delays.
This creates a competing explanation.
Perhaps utilization is falling partly because demand has changed and more care is appropriately occurring elsewhere.
Perhaps some patients are also receiving care elsewhere because the hospital has become less capable of serving them.
Both can be true at the same time.
Aurelian therefore cannot conclude that lower utilization is good.
Leadership cannot yet conclude that it is bad.
The evidence supports a narrower conclusion:
Declining utilization is a finding. Determine why it is declining before deciding what it means.
The Question Behind the Target
The investigation now returns to the original management objective:
How do we increase inpatient productivity?
Suppose leadership succeeds in moving the dashboard.
Admissions rise.
Bed-days increase.
Occupancy improves.
What would have to happen in the real world for those numbers to move?
More patients would need to be hospitalized, or hospitalized patients would need to consume additional inpatient days.
That leads to the more important question:
Do those patients actually need additional inpatient care?
If unmet demand exists because patients cannot access appropriate local hospitalization, increasing admissions may represent genuine improvement.
If patients are being appropriately managed as outpatients or referred for services unavailable locally, increasing admissions merely to raise utilization would mean something very different.
The direction of the metric cannot settle the question.
Context determines what the movement means.
Analytical Interpretation
Admissions, bed-days, and occupancy remain useful indicators.
They may reveal changing demand, excess capacity, access problems, workforce limitations, referral patterns, service changes, or shifts toward ambulatory care.
But utilization and productivity are not interchangeable concepts.
Neither is synonymous with the purpose of a healthcare system.
The dashboard answered one question:
How much inpatient capacity are we using?
Leadership was trying to answer another:
Is the hospital performing as well as it should?
The questions overlap, but they are not the same.
That distinction matters whenever a performance indicator begins to function as a target.
Before trying to move the number, leaders should ask what behavior, activity, or outcome would actually have to change to move it.
Then they can decide whether that change represents progress.
A Name for the Problem
This risk is commonly discussed through Goodhart’s law and the broader literature on performance targets.
Goodhart’s original formulation concerned the tendency of an observed statistical relationship to break down once it is subjected to pressure for control purposes.
A later and now widely used paraphrase captures the idea more simply:
When a measure becomes a target, it ceases to be a good measure.
The distinction matters because the popular sentence should not be mistaken for Goodhart’s original wording.
The broader principle, however, is useful.
When strong incentives become attached to a metric, people and organizations may change their behavior in ways that improve the measured result without necessarily improving the outcome the measure was intended to represent.
Healthcare performance literature has documented this problem: targets can improve focus and accountability while also encouraging unintended behavior or gaming.
That does not make targets inherently harmful.
It makes them something that must be interpreted with care.
The indicator is not necessarily the objective.
Beyond Hospital Beds
The same problem appears throughout complex organizations.
A hospital targets shorter length of stay. Faster discharge may reflect better patient flow—or premature discharge.
A department targets more consultations. Greater volume may mean improved access—or more activity with little additional value.
A public health program targets more screening. Higher numbers may reflect broader access—or activity poorly aligned with the population most likely to benefit.
A school rewards examination performance.
A business rewards calls completed.
A research organization rewards publications produced.
The indicator may move in the desired direction while the underlying system improves, deteriorates, or simply changes its behavior around the target.
So the useful question is not merely whether the number went up or down.
It is:
What changed in the real world to make the number move?
Key Takeaways
- Admissions and bed-days measure inpatient activity; occupancy measures capacity utilization. They should not automatically be treated as synonyms for productivity.
- Declining utilization is a finding, not an explanation.
- Demographics, clinical demand, referral patterns, institutional capability, workforce, access, and ambulatory care can all influence inpatient utilization.
- Higher utilization may represent improved access to needed care.
- Lower utilization may reflect appropriate care outside the hospital or changing population need.
- Either direction can also conceal dysfunction.
- Performance indicators provide evidence about systems; they do not automatically define what those systems should optimize.
What would have to happen in the real world to make this number move?
Is that actually the outcome we want?
A Final Thought
A performance indicator can be accurate and still be misunderstood.
The dashboard in this case may describe admissions, bed-days, and occupancy perfectly well. The harder questions begin after the measurement is complete.
Why is the indicator changing?
Does the change reflect population need, access, workforce, referral patterns, institutional capability, changes in care delivery—or several of these at once?
And if we succeed in moving the indicator, will we actually improve the outcome we care about?
That last question is the one worth protecting.
Numbers help us observe a system.
They do not decide what the system should value.
Before trying to make a number move, ask what would have to change in the real world to move it.
Then ask whether that is actually what you want.
References
- Organisation for Economic Co-operation and Development (OECD). Health at a Glance 2023: OECD Indicators. Paris: OECD Publishing; 2023. Section: Hospital beds and occupancy. doi:10.1787/7a7afb35-en.
- Organisation for Economic Co-operation and Development (OECD). Hospital Care. OECD Health. Accessed 2026. Sections addressing specialised-care consolidation, same-day care, hospital accessibility, and hospital networks.
- Ramos MC, Barreto JOM, Shimizu HE, Moraes APG, Silva EN. Regionalization for health improvement: a systematic review. PLoS ONE. 2020;15(12):e0244078. doi:10.1371/journal.pone.0244078.
- Edwards N, Black S. Targets: unintended and unanticipated effects. BMJ Quality & Safety. 2023;32(12):697–699.
- Goodhart CAE. Monetary Theory and Practice: The UK Experience. London: Macmillan; 1984. See discussion commonly known as Goodhart’s law.
Related Lessons
- Aurelian’s Casebook #1 — When the Better Hospital Looks Worse
Why a mathematically correct indicator can support the wrong conclusion when the comparison itself is inappropriate. - Field Note #7 — The Surveillance Chain: Every Signal Depends on What Happens Next
Why understanding system performance requires examining the processes connecting observation to action.
Discussion
Have you ever seen an organization try to improve a metric before determining why it was changing?
Before setting a target for an important performance indicator, what evidence would you want to see before deciding that moving the number actually represents better performance?