Hospital Intelligence Brief
How to Compare Hospital Utilization Patterns
Raw utilization differences can mislead. A fair comparison adjusts for age, case mix and geography before concluding that one hospital or region uses more care than another.
Short answer
To compare hospital utilization patterns fairly, use rates per population by age rather than raw counts, group discharges into comparable categories such as APR DRGs, adjust for severity and case mix, and compare like with like in geography and hospital type. Sources include HCUP, the Dartmouth Atlas and New York’s SPARCS public files.[1][2][3]
Why compare utilization at all?
Utilization comparisons reveal where use of hospital services differs across hospitals or regions, and therefore where a hospital might be missing demand, over-serving it or serving a different mix of patients. They also support planning: if a county’s residents are admitted far less often than the state average, the reason might be better health, better primary care, access barriers or patients traveling elsewhere. Each possibility leads to a different decision.
What measures should you use?
The most useful measures are rates and ratios. Discharges per 1,000 residents, by age group, show use independent of population size. Average length of stay shows how long patients remain. Occupancy shows how full beds are; MedPAC reported about 71 percent average occupancy in FY 2024 with 5 percent of hospitals under 13 percent and 5 percent above 90 percent.[5] Case mix and severity show how sick patients are. Emergency department visits per 1,000 and ambulatory surgery rates show outpatient use.
The Dartmouth Atlas of Health Care has long used such measures to study regional variation. It defines hospital service areas and referral regions to build comparable geographies, and its glossary defines measures like the hospital care intensity index, based on days in hospital and physician encounters.[1] Using established geographies and definitions helps ensure comparisons are fair.
How do you adjust for case mix and severity?
Hospitals treat different patients. An academic medical center treats sicker patients than a community hospital, and a comparison of raw averages will mislead. Use clinical groupers. Diagnosis-related groups classify hospital claims with similar characteristics into groups expected to have similar resource use, as the Dartmouth Atlas glossary explains.[1] The New York SPARCS public inpatient file includes the APR DRG, APR severity of illness and risk of mortality, which allows comparison within similar groups and severity levels.[3]
Compare within a DRG and severity level, or use indirect standardization: compute the number of days or cases expected given the hospital’s case mix and compare with actual. This yields ratios above or below one that are easier to interpret than raw values.
What other adjustments matter?
Age and sex structure matter for population-based rates. Hospital type matters: compare teaching hospitals with teaching hospitals, rural with rural. Payer mix matters for length of stay and service mix. Geography matters: referral patterns mean that a hospital near a boundary serves a different population than one in a core area.
The Healthcare Cost and Utilization Project offers a tool set for such work, including Clinical Classifications Software Refined, comorbidity software and hospital market structure files, built on all-payer encounter-level data.[2] Using shared tools makes your analysis easier to reproduce and defend.
How should differences be interpreted?
Variation is a prompt for questions. A hospital with a long length of stay might have sicker patients, slower discharge processes or fewer post-acute options. A region with high admission rates for conditions like asthma or heart failure might have access problems; the AHRQ Prevention Quality Indicators are designed to flag admissions for conditions where access to quality ambulatory care can reduce the likelihood of hospital care.[4] But a high rate is not proof of overuse and a low rate is not proof of quality.
Always follow quantitative comparison with local inquiry. The interactive example below uses invented numbers to illustrate how raw and adjusted comparisons can differ.
Frequently asked questions
How do you compare hospital utilization?
Use rates per population by age, group cases with clinical groupers like APR DRG, adjust for severity and compare similar hospitals and regions.
What is case-mix adjustment?
Comparing hospitals within similar diagnosis groups and severity levels, or against expected values given their case mix, so that sicker patient populations do not distort comparisons.
Where can I find utilization data?
HCUP, the Dartmouth Atlas and, for New York, SPARCS public use files.[1][2][3]
What do AHRQ Prevention Quality Indicators measure?
Admissions for conditions where access to quality ambulatory care can reduce the likelihood of hospital care.[4]
Does high utilization mean overuse?
Not necessarily. It may reflect need, access or supply; investigate before concluding.
What to do next
- Choose comparable hospitals or regions.
- Convert counts to rates by age group.
- Adjust for case mix and severity.
- Investigate the largest unexplained differences locally.
Optional · interactive example
See it in an interactive example.
Hospital Market Explorer: a market dashboard comparing a five-hospital market by service category, an outmigration capture scenario, evidence-linked answers from APIS, and a sample brief, all on synthetic data.
The experience below uses a synthetic five-hospital market built to look like public discharge data. It is not New York data, or data for any real market, and every figure is illustrative. The counts are discharges, not dollars.
- Sort the hospital table by Market share and by Discharges. Do the rankings agree?
- Select a different service category and see how each hospital’s rank changes.
- Ask APIS how our market share has changed, and where.
Explore Your Hospital’s Opportunities
Two Pillars can help hospital leadership investigate market, financial, operational and strategic questions using relevant organizational data, public information and a structured assessment process. A good first step is a market scan that uses public discharge data to show where your service area’s patients go and which gaps are worth a closer look.
Sources
- Dartmouth Atlas of Health Care, Glossary. Accessed October 2026.
- Agency for Healthcare Research and Quality, HCUP-US Overview (Healthcare Cost and Utilization Project). Accessed October 2026.
- New York State Department of Health, Hospital Inpatient Discharges (SPARCS De-Identified): 2024. Accessed October 2026.
- Agency for Healthcare Research and Quality, AHRQ Quality Indicator Tools for Data Analytics. Accessed October 2026.
- Medicare Payment Advisory Commission (MedPAC), Report to the Congress, March 2026, Chapter 3: Hospital inpatient and outpatient services. Accessed October 2026.
Sources checked October 2026. This guide is general information for healthcare leaders and is not legal, financial, clinical or compliance advice. Figures and rules are cited to the sources below and may have changed since they were accessed, so confirm anything you rely on against the current source. Two Pillars & Co is independent of the organizations cited. See our editorial policy.